Merge branch 'master' of https://github.com/odin-lang/Odin

Partial pseudo-revert
Merge pull request #2839 from Tetralux/aprintf-ally
2026-07-05 11:11:37 -07:00 · 2023-10-02 20:59:49 +01:00 · 2023-10-02 20:59:43 +01:00 · 2023-10-02 11:23:26 +02:00 · 2023-10-02 02:46:26 +00:00 · 2023-09-30 22:47:49 +02:00
791 changed files with 137596 additions and 53948 deletions
@@ -38,10 +38,10 @@ jobs:
          cd tests/vendor
          make
        timeout-minutes: 10
-      - name: Odin issues tests
+      - name: Odin internals tests
        run: |
-          cd tests/issues
-          ./run.sh
+          cd tests/internal
+          make
        timeout-minutes: 10
      - name: Odin check examples/all for Linux i386
        run: ./odin check examples/all -vet -strict-style -target:linux_i386
@@ -58,8 +58,8 @@ jobs:
      - uses: actions/checkout@v1
      - name: Download LLVM, botan and setup PATH
        run: |
-          brew install llvm@11 botan
-          echo "/usr/local/opt/llvm@11/bin" >> $GITHUB_PATH
+          brew install llvm@13 botan
+          echo "/usr/local/opt/llvm@13/bin" >> $GITHUB_PATH
          TMP_PATH=$(xcrun --show-sdk-path)/user/include
          echo "CPATH=$TMP_PATH" >> $GITHUB_ENV
      - name: build odin
@@ -92,10 +92,10 @@ jobs:
          cd tests/vendor
          make
        timeout-minutes: 10
-      - name: Odin issues tests
+      - name: Odin internals tests
        run: |
-          cd tests/issues
-          ./run.sh
+          cd tests/internal
+          make
        timeout-minutes: 10
      - name: Odin check examples/all for Darwin arm64
        run: ./odin check examples/all -vet -strict-style -target:darwin_arm64
@@ -104,13 +104,13 @@ jobs:
        run: ./odin check examples/all -vet -strict-style -target:linux_arm64
        timeout-minutes: 10
  build_windows:
-    runs-on: windows-2019
+    runs-on: windows-2022
    steps:
      - uses: actions/checkout@v1
      - name: build Odin
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          ./build.bat 1
      - name: Odin version
        run: ./odin version
@@ -121,58 +121,65 @@ jobs:
      - name: Odin check
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin check examples/demo -vet
        timeout-minutes: 10
      - name: Odin run
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin run examples/demo
        timeout-minutes: 10
      - name: Odin run -debug
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin run examples/demo -debug
        timeout-minutes: 10
      - name: Odin check examples/all
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin check examples/all -strict-style
        timeout-minutes: 10
      - name: Core library tests
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          cd tests\core
          call build.bat
        timeout-minutes: 10
      - name: Vendor library tests
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          cd tests\vendor
          call build.bat
        timeout-minutes: 10
+      - name: Odin internals tests
+        shell: cmd
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          cd tests\internal
+          call build.bat
+        timeout-minutes: 10
+      - name: Odin documentation tests
+        shell: cmd
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          cd tests\documentation
+          call build.bat
+        timeout-minutes: 10
      - name: core:math/big tests
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          cd tests\core\math\big
          call build.bat
        timeout-minutes: 10
-      - name: Odin issues tests
-        shell: cmd
-        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
-          cd tests\issues
-          call run.bat
-        timeout-minutes: 10
      - name: Odin check examples/all for Windows 32bits
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin check examples/all -strict-style -target:windows_i386
        timeout-minutes: 10
@@ -7,18 +7,19 @@ on:

 jobs:
  build_windows:
-    runs-on: windows-2019
+    if: github.repository == 'odin-lang/Odin'
+    runs-on: windows-2022
    steps:
      - uses: actions/checkout@v1
      - name: build Odin
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          ./build.bat 1 1
      - name: Odin run
        shell: cmd
        run: |
-          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat
          odin run examples/demo
      - name: Copy artifacts
        run: |
@@ -37,6 +38,7 @@ jobs:
          name: windows_artifacts
          path: dist
  build_ubuntu:
+    if: github.repository == 'odin-lang/Odin'
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v1
@@ -50,6 +52,7 @@ jobs:
        run: |
          mkdir dist
          cp odin dist
+          cp libLLVM* dist
          cp -r shared dist
          cp -r core dist
          cp -r vendor dist
@@ -60,13 +63,14 @@ jobs:
          name: ubuntu_artifacts
          path: dist
  build_macos:
+    if: github.repository == 'odin-lang/Odin'
    runs-on: macOS-latest
    steps:
      - uses: actions/checkout@v1
      - name: Download LLVM and setup PATH
        run: |
-          brew install llvm@11
-          echo "/usr/local/opt/llvm@11/bin" >> $GITHUB_PATH
+          brew install llvm@13
+          echo "/usr/local/opt/llvm@13/bin" >> $GITHUB_PATH
          TMP_PATH=$(xcrun --show-sdk-path)/user/include
          echo "CPATH=$TMP_PATH" >> $GITHUB_ENV
      - name: build odin
@@ -13,7 +13,7 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Close Stale Issues
-        uses: actions/stale@v4.1.0
+        uses: actions/stale@v7.0.0
        with:
 #          stale-issue-message: |
 #            Hello!
@@ -36,7 +36,7 @@ jobs:
 #           The motivation for this automation is to help prioritize issues in the backlog and not ignore, reject, or belittle anyone..

          days-before-stale: 120
-          days-before-close: 30
+          days-before-close: -1
          exempt-draft-pr: true
          ascending: true
          operations-per-run: 1000
@@ -22,6 +22,8 @@ bld/
 [Oo]bj/
 [Ll]og/
 ![Cc]ore/[Ll]og/
+tests/documentation/verify/
+tests/documentation/all.odin-doc
 # Visual Studio 2015 cache/options directory
 .vs/
 # Visual Studio Code options directory
@@ -268,9 +270,11 @@ bin/

 # - Linux/MacOS
 odin
+!odin/
 odin.dSYM
 *.bin
 demo.bin
+libLLVM*.so*

 # shared collection
 shared/
@@ -283,3 +287,4 @@ shared/
 *.sublime-workspace
 examples/bug/
 build.sh
+!core/debug/
@@ -11,7 +11,7 @@
        <img src="https://img.shields.io/badge/platforms-Windows%20|%20Linux%20|%20macOS-green.svg">
    </a>
    <br>
-    <a href="https://discord.gg/odinlang">
+    <a href="https://discord.com/invite/sVBPHEv">
        <img src="https://img.shields.io/discord/568138951836172421?logo=discord">
    </a>
    <a href="https://github.com/odin-lang/odin/actions">
@@ -82,7 +82,7 @@ A wiki maintained by the Odin community.

 #### [Odin Discord](https://discord.gg/sVBPHEv)

-Get live support and talk with other odiners on the Odin Discord.
+Get live support and talk with other Odin programmers on the Odin Discord.

 ### Articles

@@ -2,6 +2,23 @@

 setlocal EnableDelayedExpansion

+where /Q cl.exe || (
+	set __VSCMD_ARG_NO_LOGO=1
+	for /f "tokens=*" %%i in ('"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe" -latest -requires Microsoft.VisualStudio.Workload.NativeDesktop -property installationPath') do set VS=%%i
+	if "!VS!" equ "" (
+		echo ERROR: Visual Studio installation not found
+		exit /b 1
+	)
+	call "!VS!\VC\Auxiliary\Build\vcvarsall.bat" amd64 || exit /b 1
+)
+
+if "%VSCMD_ARG_TGT_ARCH%" neq "x64" (
+	if "%ODIN_IGNORE_MSVC_CHECK%" == "" (
+		echo ERROR: please run this from MSVC x64 native tools command prompt, 32-bit target is not supported!
+		exit /b 1
+	)
+)
+
 for /f "usebackq tokens=1,2 delims=,=- " %%i in (`wmic os get LocalDateTime /value`) do @if %%i==LocalDateTime (
 	set CURR_DATE_TIME=%%j
 )
@@ -33,8 +50,14 @@ set odin_version_raw="dev-%curr_year%-%curr_month%"
 set compiler_flags= -nologo -Oi -TP -fp:precise -Gm- -MP -FC -EHsc- -GR- -GF
 set compiler_defines= -DODIN_VERSION_RAW=\"%odin_version_raw%\"

-for /f %%i in ('git rev-parse --short HEAD') do set GIT_SHA=%%i
+if not exist .git\ goto skip_git_hash
+for /f "tokens=1,2" %%i IN ('git show "--pretty=%%cd %%h" "--date=format:%%Y-%%m" --no-patch --no-notes HEAD') do (
+	set odin_version_raw=dev-%%i
+	set GIT_SHA=%%j
+)
 if %ERRORLEVEL% equ 0 set compiler_defines=%compiler_defines% -DGIT_SHA=\"%GIT_SHA%\"
+:skip_git_hash
+
 if %nightly% equ 1 set compiler_defines=%compiler_defines% -DNIGHTLY

 if %release_mode% EQU 0 ( rem Debug
@@ -47,14 +70,26 @@ if %release_mode% EQU 0 ( rem Debug
 set compiler_warnings= ^
 	-W4 -WX ^
 	-wd4100 -wd4101 -wd4127 -wd4146 ^
+	-wd4505 ^
 	-wd4456 -wd4457

 set compiler_includes= ^
 	/Isrc\
 set libs= ^
 	kernel32.lib ^
+	Synchronization.lib ^
 	bin\llvm\windows\LLVM-C.lib

+rem DO NOT TOUCH!
+rem THIS TILDE STUFF IS FOR DEVELOPMENT ONLY!
+set tilde_backend=0
+if %tilde_backend% EQU 1 (
+	set libs=%libs% src\tilde\tb.lib
+	set compiler_defines=%compiler_defines% -DODIN_TILDE_BACKEND
+)
+rem DO NOT TOUCH!
+
+
 set linker_flags= -incremental:no -opt:ref -subsystem:console

 if %release_mode% EQU 0 ( rem Debug
@@ -6,13 +6,21 @@ set -eu
 : ${CXXFLAGS=}
 : ${LDFLAGS=}
 : ${ODIN_VERSION=dev-$(date +"%Y-%m")}
+: ${GIT_SHA=}

-CPPFLAGS="$CPPFLAGS -DODIN_VERSION_RAW=\"$ODIN_VERSION\""
 CXXFLAGS="$CXXFLAGS -std=c++14"
 LDFLAGS="$LDFLAGS -pthread -lm -lstdc++"

-GIT_SHA=$(git rev-parse --short HEAD || :)
-if [ "$GIT_SHA" ]; then CPPFLAGS="$CPPFLAGS -DGIT_SHA=\"$GIT_SHA\""; fi
+if [ -d ".git" ] && [ $(which git) ]; then
+	versionTag=( $(git show --pretty='%cd %h' --date=format:%Y-%m --no-patch --no-notes HEAD) )
+	if [ $? -eq 0 ]; then
+		ODIN_VERSION="dev-${versionTag[0]}"
+		GIT_SHA="${versionTag[1]}"
+		CPPFLAGS="$CPPFLAGS -DGIT_SHA=\"$GIT_SHA\""
+	fi
+fi
+
+CPPFLAGS="$CPPFLAGS -DODIN_VERSION_RAW=\"$ODIN_VERSION\""

 DISABLED_WARNINGS="-Wno-switch -Wno-macro-redefined -Wno-unused-value"
 OS=$(uname)
@@ -25,32 +33,50 @@ panic() {
 version() { echo "$@" | awk -F. '{ printf("%d%03d%03d%03d\n", $1,$2,$3,$4); }'; }

 config_darwin() {
-	ARCH=$(uname -m)
+	local ARCH=$(uname -m)
 	: ${LLVM_CONFIG=llvm-config}

 	# allow for arm only llvm's with version 13
-	if [ ARCH == arm64 ]; then
+	if [ "${ARCH}" == "arm64" ]; then
 		MIN_LLVM_VERSION=("13.0.0")
 	else
-		# allow for x86 / amd64 all llvm versions begining from 11
+		# allow for x86 / amd64 all llvm versions beginning from 11
 		MIN_LLVM_VERSION=("11.1.0")
 	fi

 	if [ $(version $($LLVM_CONFIG --version)) -lt $(version $MIN_LLVM_VERSION) ]; then
-		if [ ARCH == arm64 ]; then
+		if [ "${ARCH}" == "arm64" ]; then
 			panic "Requirement: llvm-config must be base version 13 for arm64"
 		else
 			panic "Requirement: llvm-config must be base version greater than 11 for amd64/x86"
 		fi
 	fi

-	LDFLAGS="$LDFLAGS -liconv -ldl"
+	MAX_LLVM_VERSION=("14.999.999")
+	if [ $(version $($LLVM_CONFIG --version)) -gt $(version $MAX_LLVM_VERSION) ]; then
+		echo "Tried to use " $(which $LLVM_CONFIG) "version" $($LLVM_CONFIG --version)
+		panic "Requirement: llvm-config must be base version smaller than 15"
+	fi
+
+	LDFLAGS="$LDFLAGS -liconv -ldl -framework System"
 	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
 	LDFLAGS="$LDFLAGS -lLLVM-C"
 }

 config_freebsd() {
-	: ${LLVM_CONFIG=/usr/local/bin/llvm-config11}
+	: ${LLVM_CONFIG=}
+
+	if [ ! "$LLVM_CONFIG" ]; then
+		if [ -x "$(command -v llvm-config11)" ]; then
+			LLVM_CONFIG=llvm-config11
+		elif [ -x "$(command -v llvm-config12)" ]; then
+			LLVM_CONFIG=llvm-config12
+		elif [ -x "$(command -v llvm-config13)" ]; then
+			LLVM_CONFIG=llvm-config13
+		else
+			panic "Unable to find LLVM-config"
+		fi
+	fi

 	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
 	LDFLAGS="$LDFLAGS $($LLVM_CONFIG --libs core native --system-libs)"
@@ -68,12 +94,14 @@ config_linux() {
 	: ${LLVM_CONFIG=}

 	if [ ! "$LLVM_CONFIG" ]; then
-		if which llvm-config > /dev/null 2>&1; then
+		if [ -x "$(command -v llvm-config)" ]; then
 			LLVM_CONFIG=llvm-config
-		elif which llvm-config-11 > /dev/null 2>&1; then
+		elif [ -x "$(command -v llvm-config-11)" ]; then
 			LLVM_CONFIG=llvm-config-11
-		elif which llvm-config-11-64 > /dev/null 2>&1; then
+		elif [ -x "$(command -v llvm-config-11-64)" ]; then
 			LLVM_CONFIG=llvm-config-11-64
+		elif [ -x "$(command -v llvm-config-14)" ]; then
+			LLVM_CONFIG=llvm-config-14
 		else
 			panic "Unable to find LLVM-config"
 		fi
@@ -85,9 +113,20 @@ config_linux() {
 		panic "Requirement: llvm-config must be base version greater than 11"
 	fi

+	MAX_LLVM_VERSION=("14.999.999")
+	if [ $(version $($LLVM_CONFIG --version)) -gt $(version $MAX_LLVM_VERSION) ]; then
+		echo "Tried to use " $(which $LLVM_CONFIG) "version" $($LLVM_CONFIG --version)
+		panic "Requirement: llvm-config must be base version smaller than 15"
+	fi
+
 	LDFLAGS="$LDFLAGS -ldl"
 	CXXFLAGS="$CXXFLAGS $($LLVM_CONFIG --cxxflags --ldflags)"
-	LDFLAGS="$LDFLAGS $($LLVM_CONFIG --libs core native --system-libs)"
+	LDFLAGS="$LDFLAGS $($LLVM_CONFIG --libs core native --system-libs --libfiles) -Wl,-rpath=\$ORIGIN"
+
+	# Creates a copy of the llvm library in the build dir, this is meant to support compiler explorer.
+	# The annoyance is that this copy can be cluttering the development folder. TODO: split staging folders
+	# for development and compiler explorer builds
+	cp $(readlink -f $($LLVM_CONFIG --libfiles)) ./
 }

 build_odin() {
@@ -99,13 +138,21 @@ build_odin() {
 		EXTRAFLAGS="-O3"
 		;;
 	release-native)
-		EXTRAFLAGS="-O3 -march=native"
+		local ARCH=$(uname -m)
+        	if [ "${ARCH}" == "arm64" ]; then
+            		# Use preferred flag for Arm (ie arm64 / aarch64 / etc)
+            		EXTRAFLAGS="-O3 -mcpu=native"
+        	else
+            		# Use preferred flag for x86 / amd64
+            		EXTRAFLAGS="-O3 -march=native"
+        	fi
 		;;
 	nightly)
 		EXTRAFLAGS="-DNIGHTLY -O3"
 		;;
 	*)
 		panic "Build mode unsupported!"
+		;;
 	esac

 	set -x
@@ -117,6 +164,14 @@ run_demo() {
 	./odin run examples/demo/demo.odin -file
 }

+have_which() {
+	if ! command -v which > /dev/null 2>&1 ; then
+		panic "Could not find \`which\`"
+	fi
+}
+
+have_which
+
 case $OS in
 Linux)
 	config_linux
@@ -132,6 +187,7 @@ FreeBSD)
 	;;
 *)
 	panic "Platform unsupported!"
+	;;
 esac

 if [[ $# -eq 0 ]]; then
@@ -15,3 +15,10 @@ if not exist "vendor\miniaudio\lib\*.lib" (
 		call build.bat
 	popd
 )
+
+
+if not exist "vendor\cgltf\lib\*.lib" (
+	pushd vendor\cgltf\src
+		call build.bat
+	popd
+)
@@ -2,25 +2,25 @@ package bufio

 import "core:io"

-// Loadahead_Reader provides io lookahead.
+// Lookahead_Reader provides io lookahead.
 // This is useful for tokenizers/parsers.
-// Loadahead_Reader is similar to bufio.Reader, but unlike bufio.Reader, Loadahead_Reader's buffer size
+// Lookahead_Reader is similar to bufio.Reader, but unlike bufio.Reader, Lookahead_Reader's buffer size
 // will EXACTLY match the specified size, whereas bufio.Reader's buffer size may differ from the specified size.
 // This makes sure that the buffer will not be accidentally read beyond the expected size.
-Loadahead_Reader :: struct {
+Lookahead_Reader :: struct {
 	r:   io.Reader,
 	buf: []byte,
 	n:   int,
 }

-lookahead_reader_init :: proc(lr: ^Loadahead_Reader, r: io.Reader, buf: []byte) -> ^Loadahead_Reader {
+lookahead_reader_init :: proc(lr: ^Lookahead_Reader, r: io.Reader, buf: []byte) -> ^Lookahead_Reader {
 	lr.r = r
 	lr.buf = buf
 	lr.n = 0
 	return lr
 }

-lookahead_reader_buffer :: proc(lr: ^Loadahead_Reader) -> []byte {
+lookahead_reader_buffer :: proc(lr: ^Lookahead_Reader) -> []byte {
 	return lr.buf[:lr.n]
 }

@@ -28,7 +28,7 @@ lookahead_reader_buffer :: proc(lr: ^Loadahead_Reader) -> []byte {
 // lookahead_reader_peek returns a slice of the Lookahead_Reader which holds n bytes
 // If the Lookahead_Reader cannot hold enough bytes, it will read from the underlying reader to populate the rest.
 // NOTE: The returned buffer is not a copy of the underlying buffer
-lookahead_reader_peek :: proc(lr: ^Loadahead_Reader, n: int) -> ([]byte, io.Error) {
+lookahead_reader_peek :: proc(lr: ^Lookahead_Reader, n: int) -> ([]byte, io.Error) {
 	switch {
 	case n < 0:
 		return nil, .Negative_Read
@@ -58,13 +58,13 @@ lookahead_reader_peek :: proc(lr: ^Loadahead_Reader, n: int) -> ([]byte, io.Erro
 // lookahead_reader_peek_all returns a slice of the Lookahead_Reader populating the full buffer
 // If the Lookahead_Reader cannot hold enough bytes, it will read from the underlying reader to populate the rest.
 // NOTE: The returned buffer is not a copy of the underlying buffer
-lookahead_reader_peek_all :: proc(lr: ^Loadahead_Reader) -> ([]byte, io.Error) {
+lookahead_reader_peek_all :: proc(lr: ^Lookahead_Reader) -> ([]byte, io.Error) {
 	return lookahead_reader_peek(lr, len(lr.buf))
 }


 // lookahead_reader_consume drops the first n populated bytes from the Lookahead_Reader.
-lookahead_reader_consume :: proc(lr: ^Loadahead_Reader, n: int) -> io.Error {
+lookahead_reader_consume :: proc(lr: ^Lookahead_Reader, n: int) -> io.Error {
 	switch {
 	case n == 0:
 		return nil
@@ -78,6 +78,6 @@ lookahead_reader_consume :: proc(lr: ^Loadahead_Reader, n: int) -> io.Error {
 	return nil
 }

-lookahead_reader_consume_all :: proc(lr: ^Loadahead_Reader) -> io.Error {
+lookahead_reader_consume_all :: proc(lr: ^Lookahead_Reader) -> io.Error {
 	return lookahead_reader_consume(lr, lr.n)
 }
@@ -14,55 +14,29 @@ read_writer_init :: proc(rw: ^Read_Writer, r: ^Reader, w: ^Writer) {
 }

 read_writer_to_stream :: proc(rw: ^Read_Writer) -> (s: io.Stream) {
-	s.stream_data = rw
-	s.stream_vtable = _read_writer_vtable
+	s.procedure = _read_writer_procedure
+	s.data = rw
 	return
 }

@(private)
-_read_writer_vtable := &io.Stream_VTable{
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_read(b, p)
-	},
-	impl_read_byte = proc(s: io.Stream) -> (c: byte, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_read_byte(b)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_unread_byte(b)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (r: rune, size: int, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_read_rune(b)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_unread_rune(b)
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).r
-		return reader_write_to(b, w)
-	},
-	impl_flush = proc(s: io.Stream)  -> io.Error {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_flush(b)
-	},
-	impl_write = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_write(b, p)
-	},
-	impl_write_byte = proc(s: io.Stream, c: byte) -> io.Error {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_write_byte(b, c)
-	},
-	impl_write_rune = proc(s: io.Stream, r: rune) -> (int, io.Error) {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_write_rune(b, r)
-	},
-	impl_read_from = proc(s: io.Stream, r: io.Reader) -> (n: i64, err: io.Error) {
-		b := (^Read_Writer)(s.stream_data).w
-		return writer_read_from(b, r)
-	},
-}
+_read_writer_procedure := proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	rw := (^Read_Writer)(stream_data)
+	n_int: int
+	#partial switch mode {
+	case .Flush:
+		err = writer_flush(rw.w)
+		return
+	case .Read:
+		n_int, err = reader_read(rw.r, p)
+		n = i64(n_int)
+		return
+	case .Write:
+		n_int, err = writer_write(rw.w, p)
+		n = i64(n_int)
+		return
+	case .Query:
+		return io.query_utility({.Flush, .Read, .Write, .Query})
+	}
+	return 0, .Empty
+}
@@ -118,9 +118,7 @@ reader_peek :: proc(b: ^Reader, n: int) -> (data: []byte, err: io.Error) {
 	b.last_rune_size = -1

 	for b.w-b.r < n && b.w-b.r < len(b.buf) && b.err == nil {
-		if fill_err := _reader_read_new_chunk(b); fill_err != nil {
-			return nil, fill_err
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	if n > len(b.buf) {
@@ -156,9 +154,7 @@ reader_discard :: proc(b: ^Reader, n: int) -> (discarded: int, err: io.Error) {
 	for {
 		skip := reader_buffered(b)
 		if skip == 0 {
-			if fill_err := _reader_read_new_chunk(b); fill_err != nil {
-				return 0, fill_err
-			}
+			_reader_read_new_chunk(b) or_return
 			skip = reader_buffered(b)
 		}
 		skip = min(skip, remaining)
@@ -223,20 +219,18 @@ reader_read :: proc(b: ^Reader, p: []byte) -> (n: int, err: io.Error) {

 // reader_read_byte reads and returns a single byte
 // If no byte is available, it return an error
-reader_read_byte :: proc(b: ^Reader) -> (byte, io.Error) {
+reader_read_byte :: proc(b: ^Reader) -> (c: byte, err: io.Error) {
 	b.last_rune_size = -1
 	for b.r == b.w {
 		if b.err != nil {
 			return 0, _reader_consume_err(b)
 		}
-		if err := _reader_read_new_chunk(b); err != nil {
-			return 0, err
-		}
+		_reader_read_new_chunk(b) or_return
 	}
-	c := b.buf[b.r]
+	c = b.buf[b.r]
 	b.r += 1
 	b.last_byte = int(c)
-	return c, nil
+	return
 }

 // reader_unread_byte unreads the last byte. Only the most recently read byte can be unread
@@ -264,15 +258,12 @@ reader_read_rune :: proc(b: ^Reader) -> (r: rune, size: int, err: io.Error) {
 	    !utf8.full_rune(b.buf[b.r:b.w]) &&
 	    b.err == nil &&
 	    b.w-b.w < len(b.buf) {
-		if err = _reader_read_new_chunk(b); err != nil {
-			return
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	b.last_rune_size = -1
 	if b.r == b.w {
-		err = _reader_consume_err(b)
-		return
+		return 0, 0, _reader_consume_err(b)
 	}
 	r, size = rune(b.buf[b.r]), 1
 	if r >= utf8.RUNE_SELF {
@@ -305,39 +296,20 @@ reader_write_to :: proc(b: ^Reader, w: io.Writer) -> (n: i64, err: io.Error) {
 		return i64(n), err
 	}

-	n, err = write_buf(b, w)
-	if err != nil {
-		return
-	}
+	n = write_buf(b, w) or_return

 	m: i64
-	if nr, ok := io.to_writer_to(b.rd); ok {
-		m, err = io.write_to(nr, w)
-		n += m
-		return n, err
-	}
-
-	if nw, ok := io.to_reader_from(w); ok {
-		m, err = io.read_from(nw, b.rd)
-		n += m
-		return n, err
-	}
-
 	if b.w-b.r < len(b.buf) {
-		if err = _reader_read_new_chunk(b); err != nil {
-			return
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	for b.r < b.w {
 		m, err = write_buf(b, w)
-		n += m
+		n += m // this needs to be done before returning
 		if err != nil {
 			return
 		}
-		if err = _reader_read_new_chunk(b); err != nil {
-			return
-		}
+		_reader_read_new_chunk(b) or_return
 	}

 	if b.err == .EOF {
@@ -352,48 +324,28 @@ reader_write_to :: proc(b: ^Reader, w: io.Writer) -> (n: i64, err: io.Error) {

 // reader_to_stream converts a Reader into an io.Stream
 reader_to_stream :: proc(b: ^Reader) -> (s: io.Stream) {
-	s.stream_data = b
-	s.stream_vtable = _reader_vtable
+	s.data = b
+	s.procedure = _reader_proc
 	return
 }



@(private)
-_reader_vtable := &io.Stream_VTable{
-	impl_destroy = proc(s: io.Stream) -> io.Error {
-		b := (^Reader)(s.stream_data)
+_reader_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	b := (^Reader)(stream_data)
+	#partial switch mode {
+	case .Read:
+		return io._i64_err(reader_read(b, p))
+	case .Destroy:
 		reader_destroy(b)
-		return nil
-	},
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_read(b, p)
-	},
-	impl_read_byte = proc(s: io.Stream) -> (c: byte, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_read_byte(b)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		b := (^Reader)(s.stream_data)
-		return reader_unread_byte(b)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (r: rune, size: int, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_read_rune(b)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		b := (^Reader)(s.stream_data)
-		return reader_unread_rune(b)
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		b := (^Reader)(s.stream_data)
-		return reader_write_to(b, w)
-	},
+		return
+	case .Query:
+		return io.query_utility({.Read, .Destroy, .Query})
+	}
+	return 0, .Empty
 }

-
-
 //
 // Utility procedures
 //
@@ -435,9 +387,7 @@ reader_read_slice :: proc(b: ^Reader, delim: byte) -> (line: []byte, err: io.Err

 		s = b.w - b.r

-		if err = _reader_read_new_chunk(b); err != nil {
-			break
-		}
+		_reader_read_new_chunk(b) or_break
 	}

 	if i := len(line)-1; i >= 0 {
@@ -66,7 +66,7 @@ scanner_destroy :: proc(s: ^Scanner) {
 }


-// Returns the first non-EOF error that was encounted by the scanner
+// Returns the first non-EOF error that was encountered by the scanner
 scanner_error :: proc(s: ^Scanner) -> Scanner_Error {
 	switch s._err {
 	case .EOF, nil:
@@ -159,7 +159,7 @@ writer_write_rune :: proc(b: ^Writer, r: rune) -> (size: int, err: io.Error) {
 	return
 }

-// writer_write writes a string into the buffer
+// writer_write_string writes a string into the buffer
 // It returns the number of bytes written
 // If n < len(p), it will return an error explaining why the write is short
 writer_write_string :: proc(b: ^Writer, s: string) -> (int, io.Error) {
@@ -173,14 +173,6 @@ writer_read_from :: proc(b: ^Writer, r: io.Reader) -> (n: i64, err: io.Error) {
 	if b.err != nil {
 		return 0, b.err
 	}
-	if writer_buffered(b) == 0 {
-		if w, ok := io.to_reader_from(b.wr); !ok {
-			n, err = io.read_from(w, r)
-			b.err = err
-			return
-		}
-	}
-
 	for {
 		if writer_available(b) == 0 {
 			writer_flush(b) or_return
@@ -222,38 +214,35 @@ writer_read_from :: proc(b: ^Writer, r: io.Reader) -> (n: i64, err: io.Error) {

 // writer_to_stream converts a Writer into an io.Stream
 writer_to_stream :: proc(b: ^Writer) -> (s: io.Stream) {
-	s.stream_data = b
-	s.stream_vtable = _writer_vtable
+	s.data = b
+	s.procedure = _writer_proc
 	return
 }

+// writer_to_stream converts a Writer into an io.Stream
+writer_to_writer :: proc(b: ^Writer) -> (s: io.Writer) {
+	return writer_to_stream(b)
+}
+


@(private)
-_writer_vtable := &io.Stream_VTable{
-	impl_destroy = proc(s: io.Stream) -> io.Error {
-		b := (^Writer)(s.stream_data)
+_writer_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	b := (^Writer)(stream_data)
+	#partial switch mode {
+	case .Flush:
+		err = writer_flush(b)
+		return
+	case .Write:
+		n_int: int
+		n_int, err = writer_write(b, p)
+		n = i64(n_int)
+		return
+	case .Destroy:
 		writer_destroy(b)
-		return nil
-	},
-	impl_flush = proc(s: io.Stream)  -> io.Error {
-		b := (^Writer)(s.stream_data)
-		return writer_flush(b)
-	},
-	impl_write = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Writer)(s.stream_data)
-		return writer_write(b, p)
-	},
-	impl_write_byte = proc(s: io.Stream, c: byte) -> io.Error {
-		b := (^Writer)(s.stream_data)
-		return writer_write_byte(b, c)
-	},
-	impl_write_rune = proc(s: io.Stream, r: rune) -> (int, io.Error) {
-		b := (^Writer)(s.stream_data)
-		return writer_write_rune(b, r)
-	},
-	impl_read_from = proc(s: io.Stream, r: io.Reader) -> (n: i64, err: io.Error) {
-		b := (^Writer)(s.stream_data)
-		return writer_read_from(b, r)
-	},
+		return
+	case .Query:
+		return io.query_utility({.Flush, .Write, .Destroy, .Query})
+	}
+	return 0, .Empty
 }
@@ -94,7 +94,15 @@ cap :: proc(array: Array_Type) -> int ---

 size_of      :: proc($T: typeid) -> int ---
 align_of     :: proc($T: typeid) -> int ---
-offset_of    :: proc($T: typeid) -> uintptr ---
+
+// e.g. offset_of(t.f), where t is an instance of the type T
+offset_of_selector :: proc(selector: $T) -> uintptr ---
+// e.g. offset_of(T, f), where T can be the type instead of a variable
+offset_of_member   :: proc($T: typeid, member: $M) -> uintptr ---
+offset_of :: proc{offset_of_selector, offset_of_member}
+// e.g. offset_of(T, "f"), where T can be the type instead of a variable
+offset_of_by_string :: proc($T: typeid, member: string) -> uintptr ---
+
 type_of      :: proc(x: expr) -> type ---
 type_info_of :: proc($T: typeid) -> ^runtime.Type_Info ---
 typeid_of    :: proc($T: typeid) -> typeid ---
@@ -109,7 +117,7 @@ jmag       :: proc(value: Quaternion) -> Float ---
 kmag       :: proc(value: Quaternion) -> Float ---
 conj       :: proc(value: Complex_Or_Quaternion) -> Complex_Or_Quaternion ---

-expand_to_tuple :: proc(value: Struct_Or_Array) -> (A, B, C, ...) ---
+expand_values :: proc(value: Struct_Or_Array) -> (A, B, C, ...) ---

 min   :: proc(values: ..T) -> T ---
 max   :: proc(values: ..T) -> T ---
@@ -38,6 +38,11 @@ buffer_init_string :: proc(b: ^Buffer, s: string) {
 }

 buffer_init_allocator :: proc(b: ^Buffer, len, cap: int, allocator := context.allocator) {
+	if b.buf == nil {
+		b.buf = make([dynamic]byte, len, cap, allocator)
+		return
+	}
+
 	b.buf.allocator = allocator
 	reserve(&b.buf, cap)
 	resize(&b.buf, len)
@@ -108,8 +113,11 @@ _buffer_grow :: proc(b: ^Buffer, n: int) -> int {
 	if i, ok := _buffer_try_grow(b, n); ok {
 		return i
 	}
+
 	if b.buf == nil && n <= SMALL_BUFFER_SIZE {
-		b.buf = make([dynamic]byte, n, SMALL_BUFFER_SIZE)
+		// Fixes #2756 by preserving allocator if already set on Buffer via init_buffer_allocator
+		reserve(&b.buf, SMALL_BUFFER_SIZE)
+		resize(&b.buf, n)
 		return 0
 	}

@@ -240,14 +248,11 @@ buffer_read_ptr :: proc(b: ^Buffer, ptr: rawptr, size: int) -> (n: int, err: io.
 buffer_read_at :: proc(b: ^Buffer, p: []byte, offset: int) -> (n: int, err: io.Error) {
 	b.last_read = .Invalid

-	if offset < 0 || offset >= len(b.buf) {
+	if uint(offset) >= len(b.buf) {
 		err = .Invalid_Offset
 		return
 	}
-
-	if 0 <= offset && offset < len(b.buf) {
-		n = copy(p, b.buf[offset:])
-	}
+	n = copy(p, b.buf[offset:])
 	if n > 0 {
 		b.last_read = .Read
 	}
@@ -373,69 +378,31 @@ buffer_read_from :: proc(b: ^Buffer, r: io.Reader) -> (n: i64, err: io.Error) #n


 buffer_to_stream :: proc(b: ^Buffer) -> (s: io.Stream) {
-	s.stream_data = b
-	s.stream_vtable = _buffer_vtable
+	s.data = b
+	s.procedure = _buffer_proc
 	return
 }

@(private)
-_buffer_vtable := &io.Stream_VTable{
-	impl_size = proc(s: io.Stream) -> i64 {
-		b := (^Buffer)(s.stream_data)
-		return i64(buffer_capacity(b))
-	},
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read(b, p)
-	},
-	impl_read_at = proc(s: io.Stream, p: []byte, offset: i64) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_at(b, p, int(offset))
-	},
-	impl_read_byte = proc(s: io.Stream) -> (byte, io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_byte(b)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (r: rune, size: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_rune(b)
-	},
-	impl_write = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write(b, p)
-	},
-	impl_write_at = proc(s: io.Stream, p: []byte, offset: i64) -> (n: int, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_at(b, p, int(offset))
-	},
-	impl_write_byte = proc(s: io.Stream, c: byte) -> io.Error {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_byte(b, c)
-	},
-	impl_write_rune = proc(s: io.Stream, r: rune) -> (int, io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_rune(b, r)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		b := (^Buffer)(s.stream_data)
-		return buffer_unread_byte(b)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		b := (^Buffer)(s.stream_data)
-		return buffer_unread_rune(b)
-	},
-	impl_destroy = proc(s: io.Stream) -> io.Error {
-		b := (^Buffer)(s.stream_data)
+_buffer_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	b := (^Buffer)(stream_data)
+	#partial switch mode {
+	case .Read:
+		return io._i64_err(buffer_read(b, p))
+	case .Read_At:
+		return io._i64_err(buffer_read_at(b, p, int(offset)))
+	case .Write:
+		return io._i64_err(buffer_write(b, p))
+	case .Write_At:
+		return io._i64_err(buffer_write_at(b, p, int(offset)))
+	case .Size:
+		n = i64(buffer_capacity(b))
+		return
+	case .Destroy:
 		buffer_destroy(b)
-		return nil
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_write_to(b, w)
-	},
-	impl_read_from = proc(s: io.Stream, r: io.Reader) -> (n: i64, err: io.Error) {
-		b := (^Buffer)(s.stream_data)
-		return buffer_read_from(b, r)
-	},
+		return
+	case .Query:
+		return io.query_utility({.Read, .Read_At, .Write, .Write_At, .Size, .Destroy})
+	}
+	return 0, .Empty
 }
-
@@ -638,7 +638,7 @@ trim_left_proc :: proc(s: []byte, p: proc(rune) -> bool) -> []byte {

 index_rune :: proc(s: []byte, r: rune) -> int {
 	switch {
-	case 0 <= r && r < utf8.RUNE_SELF:
+	case u32(r) < utf8.RUNE_SELF:
 		return index_byte(s, byte(r))

 	case r == utf8.RUNE_ERROR:
@@ -16,8 +16,8 @@ reader_init :: proc(r: ^Reader, s: []byte) {
 }

 reader_to_stream :: proc(r: ^Reader) -> (s: io.Stream) {
-	s.stream_data = r
-	s.stream_vtable = _reader_vtable
+	s.data = r
+	s.procedure = _reader_proc
 	return
 }

@@ -137,41 +137,22 @@ reader_write_to :: proc(r: ^Reader, w: io.Writer) -> (n: i64, err: io.Error) {


@(private)
-_reader_vtable := &io.Stream_VTable{
-	impl_size = proc(s: io.Stream) -> i64 {
-		r := (^Reader)(s.stream_data)
-		return reader_size(r)
-	},
-	impl_read = proc(s: io.Stream, p: []byte) -> (n: int, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read(r, p)
-	},
-	impl_read_at = proc(s: io.Stream, p: []byte, off: i64) -> (n: int, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read_at(r, p, off)
-	},
-	impl_read_byte = proc(s: io.Stream) -> (byte, io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read_byte(r)
-	},
-	impl_unread_byte = proc(s: io.Stream) -> io.Error {
-		r := (^Reader)(s.stream_data)
-		return reader_unread_byte(r)
-	},
-	impl_read_rune = proc(s: io.Stream) -> (ch: rune, size: int, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_read_rune(r)
-	},
-	impl_unread_rune = proc(s: io.Stream) -> io.Error {
-		r := (^Reader)(s.stream_data)
-		return reader_unread_rune(r)
-	},
-	impl_seek = proc(s: io.Stream, offset: i64, whence: io.Seek_From) -> (i64, io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_seek(r, offset, whence)
-	},
-	impl_write_to = proc(s: io.Stream, w: io.Writer) -> (n: i64, err: io.Error) {
-		r := (^Reader)(s.stream_data)
-		return reader_write_to(r, w)
-	},
+_reader_proc :: proc(stream_data: rawptr, mode: io.Stream_Mode, p: []byte, offset: i64, whence: io.Seek_From) -> (n: i64, err: io.Error) {
+	r := (^Reader)(stream_data)
+	#partial switch mode {
+	case .Read:
+		return io._i64_err(reader_read(r, p))
+	case .Read_At:
+		return io._i64_err(reader_read_at(r, p, offset))
+	case .Seek:
+		n, err = reader_seek(r, offset, whence)
+		return
+	case .Size:
+		n = reader_size(r)
+		return
+	case .Query:
+		return io.query_utility({.Read, .Read_At, .Seek, .Size, .Query})
+	}
+	return 0, .Empty
 }
+
@@ -1118,7 +1118,7 @@ expand_macro :: proc(cpp: ^Preprocessor, rest: ^^Token, tok: ^Token) -> bool {

 search_include_next :: proc(cpp: ^Preprocessor, filename: string) -> (path: string, ok: bool) {
 	for ; cpp.include_next_index < len(cpp.include_paths); cpp.include_next_index += 1 {
-		tpath := filepath.join(elems={cpp.include_paths[cpp.include_next_index], filename}, allocator=context.temp_allocator)
+		tpath := filepath.join({cpp.include_paths[cpp.include_next_index], filename}, allocator=context.temp_allocator)
 		if os.exists(tpath) {
 			return strings.clone(tpath), true
 		}
@@ -1136,7 +1136,7 @@ search_include_paths :: proc(cpp: ^Preprocessor, filename: string) -> (path: str
 	}

 	for include_path in cpp.include_paths {
-		tpath := filepath.join(elems={include_path, filename}, allocator=context.temp_allocator)
+		tpath := filepath.join({include_path, filename}, allocator=context.temp_allocator)
 		if os.exists(tpath) {
 			path, ok = strings.clone(tpath), true
 			cpp.filepath_cache[filename] = path
@@ -127,7 +127,7 @@ advance_rune :: proc(t: ^Tokenizer) {
 }

 advance_rune_n :: proc(t: ^Tokenizer, n: int) {
-	for in 0..<n {
+	for _ in 0..<n {
 		advance_rune(t)
 	}
 }
@@ -49,8 +49,8 @@ foreign libc {
 	// 7.3.8 Power and absolute-value functions
 	cabs    :: proc(z: complex_double) -> complex_double ---
 	cabsf   :: proc(z: complex_float) -> complex_float ---
-	cpow    :: proc(z: complex_double) -> complex_double ---
-	cpowf   :: proc(z: complex_float) -> complex_float ---
+	cpow    :: proc(x, y: complex_double) -> complex_double ---
+	cpowf   :: proc(x, y: complex_float) -> complex_float ---
 	csqrt   :: proc(z: complex_double) -> complex_double ---
 	csqrtf  :: proc(z: complex_float) -> complex_float ---

@@ -14,11 +14,24 @@ when ODIN_OS == .Windows {
 //	EDOM,
 //	EILSEQ
 //	ERANGE
-when ODIN_OS == .Linux || ODIN_OS == .FreeBSD {
+when ODIN_OS == .Linux {
 	@(private="file")
 	@(default_calling_convention="c")
 	foreign libc {
-		@(link_name="__libc_errno_location")
+		@(link_name="__errno_location")
+		_get_errno :: proc() -> ^int ---
+	}
+
+	EDOM   :: 33
+	EILSEQ :: 84
+	ERANGE :: 34
+}
+
+when ODIN_OS == .FreeBSD {
+	@(private="file")
+	@(default_calling_convention="c")
+	foreign libc {
+		@(link_name="__error")
 		_get_errno :: proc() -> ^int ---
 	}

@@ -331,7 +331,7 @@ fmin       :: proc{libc_fmin, libc_fminf}
 fma        :: proc{libc_fma, libc_fmaf}

 // But retain the 'f' suffix-variant functions as well so they can be used,
-// a trick is used here where we use explicit procedrual overloading of one
+// a trick is used here where we use explicit procedural overloading of one
 // procedure. This is done because the foreign block is marked @(private) and
 // aliasing functions does not remove privateness from the entity.
 acosf      :: proc{libc_acosf}
@@ -63,4 +63,4 @@ foreign libc {
 // strictly conformant C implementation is 16 on the platforms we care about.
 // The choice of 4096 bytes for storage of this type is more than enough on all
 // relevant platforms.
-jmp_buf :: struct #align 16 { _: [4096]char, }
+jmp_buf :: struct #align(16) { _: [4096]char, }
@@ -18,7 +18,7 @@ foreign _ {
 // strictly conformant C implementation is 16 on the platforms we care about.
 // The choice of 4096 bytes for storage of this type is more than enough on all
 // relevant platforms.
-va_list :: struct #align 16 {
+va_list :: struct #align(16) {
 	_: [4096]u8,
 }

@@ -1,7 +1,10 @@
 package libc

 when ODIN_OS == .Windows {
-	foreign import libc "system:libucrt.lib"
+	foreign import libc {
+		"system:libucrt.lib",
+		"system:legacy_stdio_definitions.lib",
+	}
 } else when ODIN_OS == .Darwin {
 	foreign import libc "system:System.framework"
 } else {
@@ -88,7 +88,6 @@ foreign libc {
 	srand         :: proc(seed: uint) ---

 	// 7.22.3 Memory management functions
-	aligned_alloc :: proc(aligment, size: size_t) -> rawptr ---
 	calloc        :: proc(nmemb, size: size_t) -> rawptr ---
 	free          :: proc(ptr: rawptr) ---
 	malloc        :: proc(size: size_t) -> rawptr ---
@@ -125,3 +124,30 @@ foreign libc {
 	mbstowcs      :: proc(pwcs: ^wchar_t, s: cstring, n: size_t) -> size_t ---
 	wcstombs      :: proc(s: [^]char, pwcs: ^wchar_t, n: size_t) -> size_t ---
 }
+
+
+aligned_alloc :: #force_inline proc "c" (alignment, size: size_t) -> rawptr {
+	when ODIN_OS == .Windows {
+		foreign libc {
+			_aligned_malloc :: proc(size, alignment: size_t) -> rawptr ---
+		}
+		return _aligned_malloc(size=size, alignment=alignment)
+	} else {
+		foreign libc {
+			aligned_alloc :: proc(alignment, size: size_t) -> rawptr ---
+		}
+		return aligned_alloc(alignment=alignment, size=size)
+	}
+}
+
+
+aligned_free :: #force_inline proc "c" (ptr: rawptr) {
+	when ODIN_OS == .Windows {
+		foreign libc {
+			_aligned_free :: proc(ptr: rawptr) ---
+		}
+		_aligned_free(ptr)
+	} else {
+		free(ptr)
+	}
+}
@@ -44,7 +44,7 @@ when ODIN_OS == .Windows {
 		@(link_name="_Cnd_destroy")   cnd_destroy   :: proc(cond: ^cnd_t) ---
 		@(link_name="_Cnd_init")      cnd_init      :: proc(cond: ^cnd_t) -> int ---
 		@(link_name="_Cnd_signal")    cnd_signal    :: proc(cond: ^cnd_t) -> int ---
-		@(link_name="_Cnd_timedwait") cnd_timedwait :: proc(cond: ^cnd_t, ts: ^timespec) -> int ---
+		@(link_name="_Cnd_timedwait") cnd_timedwait :: proc(cond: ^cnd_t, mtx: ^mtx_t, ts: ^timespec) -> int ---
 		@(link_name="_Cnd_wait")      cnd_wait      :: proc(cond: ^cnd_t, mtx: ^mtx_t) -> int ---
 		
 		// 7.26.4 Mutex functions
@@ -108,7 +108,7 @@ when ODIN_OS == .Linux {
 		cnd_destroy   :: proc(cond: ^cnd_t) ---
 		cnd_init      :: proc(cond: ^cnd_t) -> int ---
 		cnd_signal    :: proc(cond: ^cnd_t) -> int ---
-		cnd_timedwait :: proc(cond: ^cnd_t, ts: ^timespec) -> int ---
+		cnd_timedwait :: proc(cond: ^cnd_t, mtx: ^mtx_t, ts: ^timespec) -> int ---
 		cnd_wait      :: proc(cond: ^cnd_t, mtx: ^mtx_t) -> int ---
 		
 		// 7.26.4 Mutex functions
@@ -35,12 +35,12 @@ when ODIN_OS == .Windows {
 	clock_t        :: distinct long
 	time_t         :: distinct i64

-	timespec :: struct #align 8 {
+	timespec :: struct #align(8) {
 		tv_sec:  time_t,
 		tv_nsec: long,
 	}

-	tm :: struct #align 8 {
+	tm :: struct #align(8) {
 		tm_sec, tm_min, tm_hour, tm_mday, tm_mon, tm_year, tm_wday, tm_yday, tm_isdst: int,
 	}
 }
@@ -98,7 +98,7 @@ foreign libc {
 }

 // Large enough and aligned enough for any wide-spread in-use libc.
-mbstate_t :: struct #align 16 { _: [32]char, }
+mbstate_t :: struct #align(16) { _: [32]char, }

 // Odin does not have default argument promotion so the need for a separate type
 // here isn't necessary, though make it distinct just to be safe.
@@ -188,7 +188,8 @@ input_size_from_memory :: proc(z: ^Context_Memory_Input) -> (res: i64, err: Erro
 }

 input_size_from_stream :: proc(z: ^Context_Stream_Input) -> (res: i64, err: Error) {
-	return io.size(z.input), nil
+	res, _ = io.size(z.input)
+	return
 }

 input_size :: proc{input_size_from_memory, input_size_from_stream}
@@ -212,25 +213,19 @@ read_slice_from_memory :: #force_inline proc(z: ^Context_Memory_Input, size: int

@(optimization_mode="speed")
 read_slice_from_stream :: #force_inline proc(z: ^Context_Stream_Input, size: int) -> (res: []u8, err: io.Error) {
+	// TODO: REMOVE ALL USE OF context.temp_allocator here
+	// the is literally no need for it
 	b := make([]u8, size, context.temp_allocator)
-	_, e := z.input->impl_read(b[:])
-	if e == .None {
-		return b, .None
-	}
-
-	return []u8{}, e
+	_ = io.read(z.input, b[:]) or_return
+	return b, nil
 }

 read_slice :: proc{read_slice_from_memory, read_slice_from_stream}

@(optimization_mode="speed")
 read_data :: #force_inline proc(z: ^$C, $T: typeid) -> (res: T, err: io.Error) {
-	b, e := read_slice(z, size_of(T))
-	if e == .None {
-		return (^T)(&b[0])^, .None
-	}
-
-	return T{}, e
+	b := read_slice(z, size_of(T)) or_return
+	return (^T)(&b[0])^, nil
 }

@(optimization_mode="speed")
@@ -247,12 +242,8 @@ read_u8_from_memory :: #force_inline proc(z: ^Context_Memory_Input) -> (res: u8,

@(optimization_mode="speed")
 read_u8_from_stream :: #force_inline proc(z: ^Context_Stream_Input) -> (res: u8, err: io.Error) {
-	b, e := read_slice_from_stream(z, 1)
-	if e == .None {
-		return b[0], .None
-	}
-
-	return 0, e
+	b := read_slice_from_stream(z, 1) or_return
+	return b[0], nil
 }

 read_u8 :: proc{read_u8_from_memory, read_u8_from_stream}
@@ -294,17 +285,32 @@ peek_data_from_memory :: #force_inline proc(z: ^Context_Memory_Input, $T: typeid
 	}
 }

+@(optimization_mode="speed")
+peek_data_at_offset_from_memory :: #force_inline proc(z: ^Context_Memory_Input, $T: typeid, #any_int offset: int) -> (res: T, err: io.Error) {
+	size :: size_of(T)
+
+	#no_bounds_check {
+		if len(z.input_data) >= size + offset {
+			buf := z.input_data[offset:][:size]
+			return (^T)(&buf[0])^, .None
+		}
+	}
+
+	if len(z.input_data) == 0 {
+		return T{}, .EOF
+	} else {
+		return T{}, .Short_Buffer
+	}
+}
+
@(optimization_mode="speed")
 peek_data_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid) -> (res: T, err: io.Error) {
 	size :: size_of(T)

 	// Get current position to read from.
-	curr, e1 := z.input->impl_seek(0, .Current)
-	if e1 != .None {
-		return T{}, e1
-	}
-	r, e2 := io.to_reader_at(z.input)
-	if !e2 {
+	curr := z.input->impl_seek(0, .Current) or_return
+	r, e1 := io.to_reader_at(z.input)
+	if !e1 {
 		return T{}, .Empty
 	}
 	when size <= 128 {
@@ -312,8 +318,8 @@ peek_data_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid
 	} else {
 		b := make([]u8, size, context.temp_allocator)
 	}
-	_, e3 := io.read_at(r, b[:], curr)
-	if e3 != .None {
+	_, e2 := io.read_at(r, b[:], curr)
+	if e2 != .None {
 		return T{}, .Empty
 	}

@@ -321,7 +327,37 @@ peek_data_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid
 	return res, .None
 }

-peek_data :: proc{peek_data_from_memory, peek_data_from_stream}
+@(optimization_mode="speed")
+peek_data_at_offset_from_stream :: #force_inline proc(z: ^Context_Stream_Input, $T: typeid, #any_int offset: int) -> (res: T, err: io.Error) {
+	size :: size_of(T)
+
+	// Get current position to return to.
+	cur_pos := z.input->impl_seek(0, .Current) or_return
+	// Seek to offset.
+	pos := z.input->impl_seek(offset, .Start) or_return
+
+	r, e3 := io.to_reader_at(z.input)
+	if !e3 {
+		return T{}, .Empty
+	}
+	when size <= 128 {
+		b: [size]u8
+	} else {
+		b := make([]u8, size, context.temp_allocator)
+	}
+	_, e4 := io.read_at(r, b[:], pos)
+	if e4 != .None {
+		return T{}, .Empty
+	}
+
+	// Return read head to original position.
+	z.input->impl_seek(cur_pos, .Start)
+
+	res = (^T)(&b[0])^
+	return res, .None
+}
+
+peek_data :: proc{peek_data_from_memory, peek_data_from_stream, peek_data_at_offset_from_memory, peek_data_at_offset_from_stream}



@@ -407,7 +443,7 @@ peek_bits_lsb_from_memory :: #force_inline proc(z: ^Context_Memory_Input, width:
 	if z.num_bits < u64(width) {
 		refill_lsb(z)
 	}
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

@(optimization_mode="speed")
@@ -415,7 +451,7 @@ peek_bits_lsb_from_stream :: #force_inline proc(z: ^Context_Stream_Input, width:
 	if z.num_bits < u64(width) {
 		refill_lsb(z)
 	}
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

 peek_bits_lsb :: proc{peek_bits_lsb_from_memory, peek_bits_lsb_from_stream}
@@ -423,13 +459,13 @@ peek_bits_lsb :: proc{peek_bits_lsb_from_memory, peek_bits_lsb_from_stream}
@(optimization_mode="speed")
 peek_bits_no_refill_lsb_from_memory :: #force_inline proc(z: ^Context_Memory_Input, width: u8) -> u32 {
 	assert(z.num_bits >= u64(width))
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

@(optimization_mode="speed")
 peek_bits_no_refill_lsb_from_stream :: #force_inline proc(z: ^Context_Stream_Input, width: u8) -> u32 {
 	assert(z.num_bits >= u64(width))
-	return u32(z.code_buffer & ~(~u64(0) << width))
+	return u32(z.code_buffer &~ (~u64(0) << width))
 }

 peek_bits_no_refill_lsb :: proc{peek_bits_no_refill_lsb_from_memory, peek_bits_no_refill_lsb_from_stream}
@@ -1,4 +1,4 @@
-//+ignore
+//+build ignore
 package gzip

 /*
@@ -335,10 +335,8 @@ load_from_context :: proc(z: ^$C, buf: ^bytes.Buffer, known_gzip_size := -1, exp

 	// fmt.printf("GZIP: Expected Payload Size: %v\n", expected_output_size);

-	zlib_error := zlib.inflate_raw(z=z, expected_output_size=expected_output_size)
-	if zlib_error != nil {
-		return zlib_error
-	}
+	zlib.inflate_raw(z, expected_output_size=expected_output_size) or_return
+
 	/*
 		Read CRC32 using the ctx bit reader because zlib may leave bytes in there.
 	*/
@@ -177,12 +177,10 @@ decompress_slice_to_string :: proc(input: []u8, model := DEFAULT_MODEL, allocato
 	max_output_size := decompress_bound(len(input), model)

 	buf: [dynamic]u8
-	if !resize(&buf, max_output_size) {
-		return "", .Out_Of_Memory
-	}
+	resize(&buf, max_output_size) or_return

 	length, result := decompress_slice_to_output_buffer(input, buf[:])
-	resize(&buf, length)
+	resize(&buf, length) or_return
 	return string(buf[:]), result
 }
 decompress :: proc{decompress_slice_to_output_buffer, decompress_slice_to_string}
@@ -307,12 +305,10 @@ compress_string :: proc(input: string, model := DEFAULT_MODEL, allocator := cont
 	max_output_size := compress_bound(len(input))

 	buf: [dynamic]u8
-	if !resize(&buf, max_output_size) {
-		return {}, .Out_Of_Memory
-	}
+	resize(&buf, max_output_size) or_return

 	length, result := compress_string_to_buffer(input, buf[:])
-	resize(&buf, length)
+	resize(&buf, length) or_return
 	return buf[:length], result
 }
 compress :: proc{compress_string_to_buffer, compress_string}
@@ -1,4 +1,4 @@
-//+ignore
+//+build ignore
 package zlib

 /*
@@ -1,3 +1,4 @@
+//+vet !using-param
 package zlib

 /*
@@ -146,15 +147,7 @@ grow_buffer :: proc(buf: ^[dynamic]u8) -> (err: compress.Error) {
 		Double until we reach the maximum allowed.
 	*/
 	new_size := min(len(buf) << 1, compress.COMPRESS_OUTPUT_ALLOCATE_MAX)
-	resize(buf, new_size)
-	if len(buf) != new_size {
-		/*
-			Resize failed.
-		*/
-		return .Resize_Failed
-	}
-
-	return nil
+	return resize(buf, new_size)
 }

 /*
@@ -181,7 +174,7 @@ write_byte :: #force_inline proc(z: ^$C, c: u8) -> (err: io.Error) #no_bounds_ch
 }

@(optimization_mode="speed")
-repl_byte :: proc(z: ^$C, count: u16, c: u8) -> (err: io.Error) 	#no_bounds_check {
+repl_byte :: proc(z: ^$C, count: u16, c: u8) -> (err: io.Error) #no_bounds_check {
 	/*
 		TODO(Jeroen): Once we have a magic ring buffer, we can just peek/write into it
 		without having to worry about wrapping, so no need for a temp allocation to give to
@@ -305,10 +298,10 @@ decode_huffman_slowpath :: proc(z: ^$C, t: ^Huffman_Table) -> (r: u16, err: Erro
 	code := u16(compress.peek_bits_lsb(z,16))

 	k := int(z_bit_reverse(code, 16))
-	s: u8

-	#no_bounds_check for s = HUFFMAN_FAST_BITS+1; ; {
-		if k < t.maxcode[s] {
+	s: u8 = HUFFMAN_FAST_BITS+1
+	for {
+		#no_bounds_check if k < t.maxcode[s] {
 			break
 		}
 		s += 1
@@ -471,7 +464,7 @@ inflate_from_context :: proc(using ctx: ^compress.Context_Memory_Input, raw := f
 	}

 	// Parse ZLIB stream without header.
-	inflate_raw(z=ctx, expected_output_size=expected_output_size) or_return
+	inflate_raw(ctx, expected_output_size=expected_output_size) or_return

 	if !raw {
 		compress.discard_to_next_byte_lsb(ctx)
@@ -509,8 +502,8 @@ inflate_raw :: proc(z: ^$C, expected_output_size := -1, allocator := context.all
 		/*
 			Try to pre-allocate the output buffer.
 		*/
-		reserve(&z.output.buf, expected_output_size)
-		resize (&z.output.buf, expected_output_size)
+		reserve(&z.output.buf, expected_output_size) or_return
+		resize (&z.output.buf, expected_output_size) or_return
 	}

 	if len(z.output.buf) != expected_output_size {
@@ -653,7 +646,7 @@ inflate_raw :: proc(z: ^$C, expected_output_size := -1, allocator := context.all
 	}

 	if int(z.bytes_written) != len(z.output.buf) {
-		resize(&z.output.buf, int(z.bytes_written))
+		resize(&z.output.buf, int(z.bytes_written)) or_return
 	}

 	return nil
@@ -665,7 +658,7 @@ inflate_from_byte_array :: proc(input: []u8, buf: ^bytes.Buffer, raw := false, e
 	ctx.input_data = input
 	ctx.output = buf

-	return inflate_from_context(ctx=&ctx, raw=raw, expected_output_size=expected_output_size)
+	return inflate_from_context(&ctx, raw=raw, expected_output_size=expected_output_size)
 }

 inflate_from_byte_array_raw :: proc(input: []u8, buf: ^bytes.Buffer, raw := false, expected_output_size := -1) -> (err: Error) {
@@ -674,7 +667,7 @@ inflate_from_byte_array_raw :: proc(input: []u8, buf: ^bytes.Buffer, raw := fals
 	ctx.input_data = input
 	ctx.output = buf

-	return inflate_raw(z=&ctx, expected_output_size=expected_output_size)
+	return inflate_raw(&ctx, expected_output_size=expected_output_size)
 }

 inflate :: proc{inflate_from_context, inflate_from_byte_array}
@@ -27,27 +27,28 @@ Bit_Array_Iterator :: struct {
 	word_idx: int,
 	bit_idx:  uint,
 }
-
 /*
-	In:
-		- ba:   ^Bit_Array - the array to iterate over
+Wraps a `Bit_Array` into an Iterator

-	Out:
-		- it:   ^Bit_Array_Iterator - the iterator that holds iteration state
+Inputs:
+- ba: Pointer to the Bit_Array
+
+Returns:
+- it: Iterator struct
 */
 make_iterator :: proc (ba: ^Bit_Array) -> (it: Bit_Array_Iterator) {
 	return Bit_Array_Iterator { array = ba }
 }
-
 /*
-	In:
-		- it:    ^Bit_Array_Iterator - the iterator struct that holds the state.
+Returns the next bit, including its set-state. ok=false once exhausted

-	Out:
-		- set:    bool - the state of the bit at `index`
-		- index:  int - the next bit of the Bit_Array referenced by `it`.
-		- ok:	  bool - `true` if the iterator returned a valid index,
-			  `false` if there were no more bits
+Inputs:
+- it: The iterator that holds the state.
+
+Returns:
+- set: `true` if the bit at `index` is set.
+- index: The next bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
 */
 iterate_by_all :: proc (it: ^Bit_Array_Iterator) -> (set: bool, index: int, ok: bool) {
 	index = it.word_idx * NUM_BITS + int(it.bit_idx) + it.array.bias
@@ -64,39 +65,51 @@ iterate_by_all :: proc (it: ^Bit_Array_Iterator) -> (set: bool, index: int, ok:

 	return set, index, true
 }
-
 /*
-	In:
-		- it:     ^Bit_Array_Iterator - the iterator struct that holds the state.
+Returns the next Set Bit, for example if `0b1010`, then the iterator will return index={1, 3} over two calls.

-	Out:
-		- index:  int - the next set bit of the Bit_Array referenced by `it`.
-		- ok:	  bool - `true` if the iterator returned a valid index,
-			  `false` if there were no more bits set
+Inputs:
+- it: The iterator that holds the state.
+
+Returns:
+- index: The next *set* bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
 */
 iterate_by_set :: proc (it: ^Bit_Array_Iterator) -> (index: int, ok: bool) {
 	return iterate_internal_(it, true)
 }
-
 /*
-	In:
-		- it:	  ^Bit_Array_Iterator - the iterator struct that holds the state.
+Returns the next Unset Bit, for example if `0b1010`, then the iterator will return index={0, 2} over two calls.

-	Out:
-		- index:  int - the next unset bit of the Bit_Array referenced by `it`.
-		- ok:	  bool - `true` if the iterator returned a valid index,
-			  `false` if there were no more unset bits
+Inputs:
+- it: The iterator that holds the state.
+
+Returns:
+- index: The next *unset* bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
 */
 iterate_by_unset:: proc (it: ^Bit_Array_Iterator) -> (index: int, ok: bool) {
 	return iterate_internal_(it, false)
 }
+/*
+Iterates through set/unset bits

+*Private*
+
+Inputs:
+- it: The iterator that holds the state.
+- ITERATE_SET_BITS: `true` for returning only set bits, false for returning only unset bits
+
+Returns:
+- index: The next *unset* bit of the Bit_Array referenced by `it`.
+- ok: `true` if the iterator can continue, `false` if the iterator is done
+*/
@(private="file")
 iterate_internal_ :: proc (it: ^Bit_Array_Iterator, $ITERATE_SET_BITS: bool) -> (index: int, ok: bool) {
 	word := it.array.bits[it.word_idx] if len(it.array.bits) > it.word_idx else 0
 	when ! ITERATE_SET_BITS { word = ~word }

-	// if the word is empty or we have already gone over all the bits in it,
+	// If the word is empty or we have already gone over all the bits in it,
 	// b.bit_idx is greater than the index of any set bit in the word,
 	// meaning that word >> b.bit_idx == 0.
 	for it.word_idx < len(it.array.bits) && word >> it.bit_idx == 0 {
@@ -106,14 +119,14 @@ iterate_internal_ :: proc (it: ^Bit_Array_Iterator, $ITERATE_SET_BITS: bool) ->
 		when ! ITERATE_SET_BITS { word = ~word }
 	}

-	// if we are iterating the set bits, reaching the end of the array means we have no more bits to check
+	// If we are iterating the set bits, reaching the end of the array means we have no more bits to check
 	when ITERATE_SET_BITS {
 		if it.word_idx >= len(it.array.bits) {
 			return 0, false
 		}
 	}

-	// reaching here means that the word has some set bits
+	// Reaching here means that the word has some set bits
 	it.bit_idx += uint(intrinsics.count_trailing_zeros(word >> it.bit_idx))
 	index = it.word_idx * NUM_BITS + int(it.bit_idx) + it.array.bias

@@ -124,24 +137,21 @@ iterate_internal_ :: proc (it: ^Bit_Array_Iterator, $ITERATE_SET_BITS: bool) ->
 	}
 	return index, index <= it.array.max_index
 }
-
-
 /*
-	In:
-		- ba:    ^Bit_Array - a pointer to the Bit Array
-		- index: The bit index. Can be an enum member.
+Gets the state of a bit in the bit-array

-	Out:
-		- res:   The bit you're interested in.
-		- ok:    Whether the index was valid. Returns `false` if the index is smaller than the bias.
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array

-	The `ok` return value may be ignored.
+Returns:
+- res: `true` if the bit at `index` is set.
+- ok: Whether the index was valid. Returns `false` if the index is smaller than the bias.
 */
-get :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (res: bool, ok: bool) {
+get :: proc(ba: ^Bit_Array, #any_int index: uint) -> (res: bool, ok: bool) #optional_ok {
 	idx := int(index) - ba.bias

 	if ba == nil || int(index) < ba.bias { return false, false }
-	context.allocator = allocator

 	leg_index := idx >> INDEX_SHIFT
 	bit_index := idx &  INDEX_MASK
@@ -157,18 +167,36 @@ get :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator

 	return res, true
 }
-
 /*
-	In:
-		- ba:    ^Bit_Array - a pointer to the Bit Array
-		- index: The bit index. Can be an enum member.
+Gets the state of a bit in the bit-array

-	Out:
-		- ok:    Whether or not we managed to set requested bit.
+*Bypasses all Checks*

-	`set` automatically resizes the Bit Array to accommodate the requested index if needed.
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+
+Returns:
+- `true` if bit is set
 */
-set :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (ok: bool) {
+unsafe_get :: #force_inline proc(ba: ^Bit_Array, #any_int index: uint) -> bool #no_bounds_check {
+	return bool((ba.bits[index >> INDEX_SHIFT] >> uint(index & INDEX_MASK)) & 1)
+}
+/*
+Sets the state of a bit in the bit-array
+
+*Conditionally Allocates (Resizes backing data when `index > len(ba.bits)`)*
+
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+- set_to: `true` sets the bit on, `false` to turn it off
+- allocator: (default is context.allocator)
+
+Returns:
+- ok: Whether the set was successful, `false` on allocation failure or bad index
+*/
+set :: proc(ba: ^Bit_Array, #any_int index: uint, set_to: bool = true, allocator := context.allocator) -> (ok: bool) {

 	idx := int(index) - ba.bias

@@ -181,65 +209,97 @@ set :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator
 	resize_if_needed(ba, leg_index) or_return

 	ba.max_index = max(idx, ba.max_index)
-	ba.bits[leg_index] |= 1 << uint(bit_index)
+
+	if set_to{ ba.bits[leg_index] |= 1 << uint(bit_index) }
+	else { ba.bits[leg_index] &= ~(1 << uint(bit_index)) }
+
 	return true
 }
-
 /*
-	In:
-		- ba:    ^Bit_Array - a pointer to the Bit Array
-		- index: The bit index. Can be an enum member.
+Sets the state of a bit in the bit-array

-	Out:
-		- ok:    Whether or not we managed to unset requested bit.
+*Bypasses all checks*

-	`unset` automatically resizes the Bit Array to accommodate the requested index if needed.
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
 */
-unset :: proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (ok: bool) {
-
-	idx := int(index) - ba.bias
-
-	if ba == nil || int(index) < ba.bias { return false }
-	context.allocator = allocator
-
-	leg_index := idx >> INDEX_SHIFT
-	bit_index := idx &  INDEX_MASK
-
-	resize_if_needed(ba, leg_index) or_return
-
-	ba.max_index = max(idx, ba.max_index)
-	ba.bits[leg_index] &= ~(1 << uint(bit_index))
-	return true
+unsafe_set :: proc(ba: ^Bit_Array, bit: int) #no_bounds_check {
+	ba.bits[bit >> INDEX_SHIFT] |= 1 << uint(bit & INDEX_MASK)
 }
-
 /*
-	A helper function to create a Bit Array with optional bias, in case your smallest index is non-zero (including negative).
+Unsets the state of a bit in the bit-array. (Convienence wrapper for `set`)
+
+*Conditionally Allocates (Resizes backing data when `index > len(ba.bits)`)*
+
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+- allocator: (default is context.allocator)
+
+Returns:
+- ok: Whether the unset was successful, `false` on allocation failure or bad index
 */
-create :: proc(max_index: int, min_index := 0, allocator := context.allocator) -> (res: ^Bit_Array, ok: bool) #optional_ok {
+unset :: #force_inline proc(ba: ^Bit_Array, #any_int index: uint, allocator := context.allocator) -> (ok: bool) {
+	return set(ba, index, false, allocator)
+}
+/*
+Unsets the state of a bit in the bit-array
+
+*Bypasses all Checks*
+
+Inputs:
+- ba: Pointer to the Bit_Array
+- index: Which bit in the array
+*/
+unsafe_unset :: proc(b: ^Bit_Array, bit: int) #no_bounds_check {
+	b.bits[bit >> INDEX_SHIFT] &= ~(1 << uint(bit & INDEX_MASK))
+}
+/*
+A helper function to create a Bit Array with optional bias, in case your smallest index is non-zero (including negative).
+
+*Allocates (`new(Bit_Array) & make(ba.bits)`)*
+
+Inputs:
+- max_index: maximum starting index
+- min_index: minimum starting index (used as a bias)
+- allocator: (default is context.allocator)
+
+Returns:
+- ba: Allocates a bit_Array, backing data is set to `max-min / 64` indices, rounded up (eg 65 - 0 allocates for [2]u64).
+*/
+create :: proc(max_index: int, min_index: int = 0, allocator := context.allocator) -> (res: ^Bit_Array, ok: bool) #optional_ok {
 	context.allocator = allocator
 	size_in_bits := max_index - min_index

 	if size_in_bits < 1 { return {}, false }

 	legs := size_in_bits >> INDEX_SHIFT
-
+	if size_in_bits & INDEX_MASK > 0 {legs+=1}
+	bits, err := make([dynamic]u64, legs)
+	ok = err == mem.Allocator_Error.None
 	res = new(Bit_Array)
+	res.bits         = bits
 	res.bias         = min_index
 	res.max_index    = max_index
 	res.free_pointer = true
-	return res, resize_if_needed(res, legs)
+	return
 }
-
 /*
-	Sets all bits to `false`.
+Sets all values in the Bit_Array to zero.
+
+Inputs:
+- ba: The target Bit_Array
 */
 clear :: proc(ba: ^Bit_Array) {
 	if ba == nil { return }
 	mem.zero_slice(ba.bits[:])
 }
-
 /*
-	Releases the memory used by the Bit Array.
+Deallocates the Bit_Array and its backing storage
+
+Inputs:
+- ba: The target Bit_Array
 */
 destroy :: proc(ba: ^Bit_Array) {
 	if ba == nil { return }
@@ -248,9 +308,8 @@ destroy :: proc(ba: ^Bit_Array) {
 		free(ba)
 	}
 }
-
 /*
-	Resizes the Bit Array. For internal use.
+	Resizes the Bit Array. For internal use. Provisions needed capacity+1
 	If you want to reserve the memory for a given-sized Bit Array up front, you can use `create`.
 */
@(private="file")
@@ -20,10 +20,10 @@ List :: struct {


 Node :: struct {
-	next, prev: ^Node,
+	prev, next: ^Node,
 }

-push_front :: proc(list: ^List, node: ^Node) {
+push_front :: proc "contextless" (list: ^List, node: ^Node) {
 	if list.head != nil {
 		list.head.prev = node
 		node.prev, node.next = nil, list.head
@@ -34,7 +34,7 @@ push_front :: proc(list: ^List, node: ^Node) {
 	}
 }

-push_back :: proc(list: ^List, node: ^Node) {
+push_back :: proc "contextless" (list: ^List, node: ^Node) {
 	if list.tail != nil {
 		list.tail.next = node
 		node.prev, node.next = list.tail, nil
@@ -45,7 +45,7 @@ push_back :: proc(list: ^List, node: ^Node) {
 	}
 }

-remove :: proc(list: ^List, node: ^Node) {
+remove :: proc "contextless" (list: ^List, node: ^Node) {
 	if node != nil {
 		if node.next != nil {
 			node.next.prev = node.prev
@@ -83,12 +83,34 @@ remove_by_proc :: proc(list: ^List, to_erase: proc(^Node) -> bool) {
 	}
 }

+remove_by_proc_contextless :: proc(list: ^List, to_erase: proc "contextless" (^Node) -> bool) {
+	for node := list.head; node != nil; {
+		next := node.next
+		if to_erase(node) {
+			if node.next != nil {
+				node.next.prev = node.prev
+			}
+			if node.prev != nil {
+				node.prev.next = node.next
+			}
+			if list.head == node {
+				list.head = node.next
+			}
+			if list.tail == node {
+				list.tail = node.prev
+			}
+		}
+		node = next
+	}
+}

-is_empty :: proc(list: ^List) -> bool {
+
+
+is_empty :: proc "contextless" (list: ^List) -> bool {
 	return list.head == nil
 }

-pop_front :: proc(list: ^List) -> ^Node {
+pop_front :: proc "contextless" (list: ^List) -> ^Node {
 	link := list.head
 	if link == nil {
 		return nil
@@ -108,7 +130,7 @@ pop_front :: proc(list: ^List) -> ^Node {
 	return link

 }
-pop_back :: proc(list: ^List) -> ^Node {
+pop_back :: proc "contextless" (list: ^List) -> ^Node {
 	link := list.tail
 	if link == nil {
 		return nil
@@ -134,25 +156,25 @@ Iterator :: struct($T: typeid) {
 	offset: uintptr,
 }

-iterator_head :: proc(list: List, $T: typeid, $field_name: string) -> Iterator(T)
+iterator_head :: proc "contextless" (list: List, $T: typeid, $field_name: string) -> Iterator(T)
 	where intrinsics.type_has_field(T, field_name),
 	      intrinsics.type_field_type(T, field_name) == Node {
 	return {list.head, offset_of_by_string(T, field_name)}
 }

-iterator_tail :: proc(list: List, $T: typeid, $field_name: string) -> Iterator(T)
+iterator_tail :: proc "contextless" (list: List, $T: typeid, $field_name: string) -> Iterator(T)
 	where intrinsics.type_has_field(T, field_name),
 	      intrinsics.type_field_type(T, field_name) == Node {
 	return {list.tail, offset_of_by_string(T, field_name)}
 }

-iterator_from_node :: proc(node: ^Node, $T: typeid, $field_name: string) -> Iterator(T)
+iterator_from_node :: proc "contextless" (node: ^Node, $T: typeid, $field_name: string) -> Iterator(T)
 	where intrinsics.type_has_field(T, field_name),
 	      intrinsics.type_field_type(T, field_name) == Node {
 	return {node, offset_of_by_string(T, field_name)}
 }

-iterate_next :: proc(it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
+iterate_next :: proc "contextless" (it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
 	node := it.curr
 	if node == nil {
 		return nil, false
@@ -162,7 +184,7 @@ iterate_next :: proc(it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
 	return (^T)(uintptr(node) - it.offset), true
 }

-iterate_prev :: proc(it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
+iterate_prev :: proc "contextless" (it: ^Iterator($T)) -> (ptr: ^T, ok: bool) {
 	node := it.curr
 	if node == nil {
 		return nil, false
@@ -85,7 +85,6 @@ _shift_down :: proc(pq: ^$Q/Priority_Queue($T), i0, n: int) -> bool {
 _shift_up :: proc(pq: ^$Q/Priority_Queue($T), j: int) {
 	j := j
 	queue := pq.queue[:]
-	n := builtin.len(queue)
 	for 0 <= j {
 		i := (j-1)/2
 		if i == j || !pq.less(queue[j], queue[i]) {
@@ -14,7 +14,7 @@ Queue :: struct($T: typeid) {
 DEFAULT_CAPACITY :: 16

 // Procedure to initialize a queue
-init :: proc(q: ^$Q/Queue($T), capacity := DEFAULT_CAPACITY, allocator := context.allocator) -> bool {
+init :: proc(q: ^$Q/Queue($T), capacity := DEFAULT_CAPACITY, allocator := context.allocator) -> runtime.Allocator_Error {
 	if q.data.allocator.procedure == nil {
 		q.data.allocator = allocator
 	}
@@ -55,11 +55,11 @@ space :: proc(q: $Q/Queue($T)) -> int {
 }

 // Reserve enough space for at least the specified capacity
-reserve :: proc(q: ^$Q/Queue($T), capacity: int) -> bool {
-	if uint(capacity) > q.len {
+reserve :: proc(q: ^$Q/Queue($T), capacity: int) -> runtime.Allocator_Error {
+	if capacity > space(q^) {
 		return _grow(q, uint(capacity)) 
 	}
-	return true
+	return nil
 }


@@ -73,11 +73,18 @@ get :: proc(q: ^$Q/Queue($T), #any_int i: int, loc := #caller_location) -> T {
 front :: proc(q: ^$Q/Queue($T)) -> T {
 	return q.data[q.offset]
 }
+front_ptr :: proc(q: ^$Q/Queue($T)) -> ^T {
+	return &q.data[q.offset]
+}

 back :: proc(q: ^$Q/Queue($T)) -> T {
 	idx := (q.offset+uint(q.len))%builtin.len(q.data)
 	return q.data[idx]
 }
+back_ptr :: proc(q: ^$Q/Queue($T)) -> ^T {
+	idx := (q.offset+uint(q.len))%builtin.len(q.data)
+	return &q.data[idx]
+}

 set :: proc(q: ^$Q/Queue($T), #any_int i: int, val: T, loc := #caller_location) {
 	runtime.bounds_check_error_loc(loc, i, builtin.len(q.data))
@@ -92,26 +99,38 @@ get_ptr :: proc(q: ^$Q/Queue($T), #any_int i: int, loc := #caller_location) -> ^
 	return &q.data[idx]
 }

+peek_front :: proc(q: ^$Q/Queue($T), loc := #caller_location) -> ^T {
+	runtime.bounds_check_error_loc(loc, 0, builtin.len(q.data))
+	idx := q.offset%builtin.len(q.data)
+	return &q.data[idx]
+}
+
+peek_back :: proc(q: ^$Q/Queue($T), loc := #caller_location) -> ^T {
+	runtime.bounds_check_error_loc(loc, int(q.len - 1), builtin.len(q.data))
+	idx := (uint(q.len - 1)+q.offset)%builtin.len(q.data)
+	return &q.data[idx]
+}
+
 // Push an element to the back of the queue
-push_back :: proc(q: ^$Q/Queue($T), elem: T) -> bool {
+push_back :: proc(q: ^$Q/Queue($T), elem: T) -> (ok: bool, err: runtime.Allocator_Error) {
 	if space(q^) == 0 {
 		_grow(q) or_return
 	}
 	idx := (q.offset+uint(q.len))%builtin.len(q.data)
 	q.data[idx] = elem
 	q.len += 1
-	return true
+	return true, nil
 }

 // Push an element to the front of the queue
-push_front :: proc(q: ^$Q/Queue($T), elem: T) -> bool {
+push_front :: proc(q: ^$Q/Queue($T), elem: T) -> (ok: bool, err: runtime.Allocator_Error)  {
 	if space(q^) == 0 {
 		_grow(q) or_return
 	}	
 	q.offset = uint(q.offset - 1 + builtin.len(q.data)) % builtin.len(q.data)
 	q.len += 1
 	q.data[q.offset] = elem
-	return true
+	return true, nil
 }


@@ -154,7 +173,7 @@ pop_front_safe :: proc(q: ^$Q/Queue($T)) -> (elem: T, ok: bool) {
 }

 // Push multiple elements to the front of the queue
-push_back_elems :: proc(q: ^$Q/Queue($T), elems: ..T) -> bool {
+push_back_elems :: proc(q: ^$Q/Queue($T), elems: ..T) -> (ok: bool, err: runtime.Allocator_Error)  {
 	n := uint(builtin.len(elems))
 	if space(q^) < int(n) {
 		_grow(q, q.len + n) or_return
@@ -169,7 +188,7 @@ push_back_elems :: proc(q: ^$Q/Queue($T), elems: ..T) -> bool {
 	copy(q.data[insert_from:], elems[:insert_to])
 	copy(q.data[:insert_from], elems[insert_to:])
 	q.len += n
-	return true
+	return true, nil
 }

 // Consume `n` elements from the front of the queue
@@ -206,7 +225,7 @@ clear :: proc(q: ^$Q/Queue($T)) {


 // Internal growinh procedure
-_grow :: proc(q: ^$Q/Queue($T), min_capacity: uint = 0) -> bool {
+_grow :: proc(q: ^$Q/Queue($T), min_capacity: uint = 0) -> runtime.Allocator_Error {
 	new_capacity := max(min_capacity, uint(8), uint(builtin.len(q.data))*2)
 	n := uint(builtin.len(q.data))
 	builtin.resize(&q.data, int(new_capacity)) or_return
@@ -215,5 +234,5 @@ _grow :: proc(q: ^$Q/Queue($T), min_capacity: uint = 0) -> bool {
 		copy(q.data[new_capacity-diff:], q.data[q.offset:][:diff])
 		q.offset += new_capacity - n
 	}
-	return true
+	return nil
 }
@@ -1,6 +1,8 @@
 package container_small_array

 import "core:builtin"
+import "core:runtime"
+_ :: runtime

 Small_Array :: struct($N: int, $T: typeid) where N >= 0 {
 	data: [N]T,
@@ -8,40 +10,54 @@ Small_Array :: struct($N: int, $T: typeid) where N >= 0 {
 }


-len :: proc(a: $A/Small_Array) -> int {
+len :: proc "contextless" (a: $A/Small_Array) -> int {
 	return a.len
 }

-cap :: proc(a: $A/Small_Array) -> int {
+cap :: proc "contextless" (a: $A/Small_Array) -> int {
 	return builtin.len(a.data)
 }

-space :: proc(a: $A/Small_Array) -> int {
+space :: proc "contextless" (a: $A/Small_Array) -> int {
 	return builtin.len(a.data) - a.len
 }

-slice :: proc(a: ^$A/Small_Array($N, $T)) -> []T {
+slice :: proc "contextless" (a: ^$A/Small_Array($N, $T)) -> []T {
 	return a.data[:a.len]
 }


-get :: proc(a: $A/Small_Array($N, $T), index: int) -> T {
+get :: proc "contextless" (a: $A/Small_Array($N, $T), index: int) -> T {
 	return a.data[index]
 }
-get_ptr :: proc(a: ^$A/Small_Array($N, $T), index: int) -> ^T {
+get_ptr :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int) -> ^T {
 	return &a.data[index]
 }

-set :: proc(a: ^$A/Small_Array($N, $T), index: int, item: T) {
+get_safe :: proc(a: $A/Small_Array($N, $T), index: int) -> (T, bool) #no_bounds_check {
+	if index < 0 || index >= a.len {
+		return {}, false
+	}
+	return a.data[index], true
+}
+
+get_ptr_safe :: proc(a: ^$A/Small_Array($N, $T), index: int) -> (^T, bool) #no_bounds_check {
+	if index < 0 || index >= a.len {
+		return {}, false
+	}
+	return &a.data[index], true
+}
+
+set :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int, item: T) {
 	a.data[index] = item
 }

-resize :: proc(a: ^$A/Small_Array, length: int) {
+resize :: proc "contextless" (a: ^$A/Small_Array, length: int) {
 	a.len = min(length, builtin.len(a.data))
 }


-push_back :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
+push_back :: proc "contextless" (a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	if a.len < cap(a^) {
 		a.data[a.len] = item
 		a.len += 1
@@ -50,7 +66,7 @@ push_back :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	return false
 }

-push_front :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
+push_front :: proc "contextless" (a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	if a.len < cap(a^) {
 		a.len += 1
 		data := slice(a)
@@ -61,14 +77,14 @@ push_front :: proc(a: ^$A/Small_Array($N, $T), item: T) -> bool {
 	return false
 }

-pop_back :: proc(a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
+pop_back :: proc "odin" (a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
 	assert(condition=(N > 0 && a.len > 0), loc=loc)
 	item := a.data[a.len-1]
 	a.len -= 1
 	return item
 }

-pop_front :: proc(a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
+pop_front :: proc "odin" (a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
 	assert(condition=(N > 0 && a.len > 0), loc=loc)
 	item := a.data[0]
 	s := slice(a)
@@ -77,7 +93,7 @@ pop_front :: proc(a: ^$A/Small_Array($N, $T), loc := #caller_location) -> T {
 	return item
 }

-pop_back_safe :: proc(a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
+pop_back_safe :: proc "contextless" (a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
 	if N > 0 && a.len > 0 {
 		item = a.data[a.len-1]
 		a.len -= 1
@@ -86,31 +102,60 @@ pop_back_safe :: proc(a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
 	return
 }

-pop_front_safe :: proc(a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
+pop_front_safe :: proc "contextless" (a: ^$A/Small_Array($N, $T)) -> (item: T, ok: bool) {
 	if N > 0 && a.len > 0 {
 		item = a.data[0]
 		s := slice(a)
 		copy(s[:], s[1:])
 		a.len -= 1
 		ok = true
-	} 
+	}
 	return
 }

-consume :: proc(a: ^$A/Small_Array($N, $T), count: int, loc := #caller_location) {
+consume :: proc "odin" (a: ^$A/Small_Array($N, $T), count: int, loc := #caller_location) {
 	assert(condition=a.len >= count, loc=loc)
 	a.len -= count
 }

-clear :: proc(a: ^$A/Small_Array($N, $T)) {
+ordered_remove :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int, loc := #caller_location) #no_bounds_check {
+    runtime.bounds_check_error_loc(loc, index, a.len)
+    if index+1 < a.len {
+		copy(a.data[index:], a.data[index+1:])
+	}
+	a.len -= 1
+}
+
+unordered_remove :: proc "contextless" (a: ^$A/Small_Array($N, $T), index: int, loc := #caller_location) #no_bounds_check {
+    runtime.bounds_check_error_loc(loc, index, a.len)
+	n := a.len-1
+	if index != n {
+		a.data[index] = a.data[n]
+	}
+    a.len -= 1
+}
+
+clear :: proc "contextless" (a: ^$A/Small_Array($N, $T)) {
 	resize(a, 0)
 }

-push_back_elems :: proc(a: ^$A/Small_Array($N, $T), items: ..T) {
+push_back_elems :: proc "contextless" (a: ^$A/Small_Array($N, $T), items: ..T) {
 	n := copy(a.data[a.len:], items[:])
 	a.len += n
 }

+inject_at :: proc "contextless" (a: ^$A/Small_Array($N, $T), item: T, index: int) -> bool #no_bounds_check {
+	if a.len < cap(a^) && index >= 0 && index <= len(a^) {
+		a.len += 1
+		for i := a.len - 1; i >= index + 1; i -= 1 {
+			a.data[i] = a.data[i - 1]
+		}
+		a.data[index] = item
+		return true
+	}
+	return false
+}
+
 append_elem  :: push_back
 append_elems :: push_back_elems
 push   :: proc{push_back, push_back_elems}
@@ -32,7 +32,7 @@ init :: proc(sorter: ^$S/Sorter($K)) {
 }

 destroy :: proc(sorter: ^$S/Sorter($K)) {
-	for _, v in &sorter.relations {
+	for _, v in sorter.relations {
 		delete(v.dependents)
 	}
 	delete(sorter.relations)
@@ -80,7 +80,7 @@ sort :: proc(sorter: ^$S/Sorter($K)) -> (sorted, cycled: [dynamic]K) {
 		}
 	}

-	for root in &sorted do for k, _ in relations[root].dependents {
+	for root in sorted do for k, _ in relations[root].dependents {
 		relation := &relations[k]
 		relation.dependencies -= 1
 		if relation.dependencies == 0 {
@@ -81,7 +81,7 @@ The crypto package is not thread-safe at the moment. This may change in the futu
 ### Disclaimer
 The algorithms were ported out of curiosity and due to interest in the field.
 We have not had any of the code verified by a third party or tested/fuzzed by any automatic means.
-Whereever we were able to find official test vectors, those were used to verify the implementation.
+Wherever we were able to find official test vectors, those were used to verify the implementation.
 We do not recommend using them in a production environment, without any additional testing and/or verification.

 ### ToDo
@@ -30,6 +30,6 @@ equivalence.

 For the most part, alterations to the base fiat-crypto generated code was
 kept to a minimum, to aid auditability.  This results in a somewhat
-ideosyncratic style, and in some cases minor performance penalties.
+idiosyncratic style, and in some cases minor performance penalties.

 [1]: https://github.com/mit-plv/fiat-crypto
@@ -9,14 +9,16 @@ package fiat
 u1 :: distinct u8
 i1 :: distinct i8

-cmovznz_u64 :: #force_inline proc "contextless" (arg1: u1, arg2, arg3: u64) -> (out1: u64) {
+@(optimization_mode="none")
+cmovznz_u64 :: proc "contextless" (arg1: u1, arg2, arg3: u64) -> (out1: u64) {
 	x1 := (u64(arg1) * 0xffffffffffffffff)
 	x2 := ((x1 & arg3) | ((~x1) & arg2))
 	out1 = x2
 	return
 }

-cmovznz_u32 :: #force_inline proc "contextless" (arg1: u1, arg2, arg3: u32) -> (out1: u32) {
+@(optimization_mode="none")
+cmovznz_u32 :: proc "contextless" (arg1: u1, arg2, arg3: u32) -> (out1: u32) {
 	x1 := (u32(arg1) * 0xffffffff)
 	x2 := ((x1 & arg3) | ((~x1) & arg2))
 	out1 = x2
@@ -305,7 +305,8 @@ fe_opp :: proc "contextless" (out1: ^Loose_Field_Element, arg1: ^Tight_Field_Ele
 	out1[4] = x5
 }

-fe_cond_assign :: proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: int) {
+@(optimization_mode="none")
+fe_cond_assign :: #force_no_inline proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: int) {
 	x1 := fiat.cmovznz_u64(fiat.u1(arg2), out1[0], arg1[0])
 	x2 := fiat.cmovznz_u64(fiat.u1(arg2), out1[1], arg1[1])
 	x3 := fiat.cmovznz_u64(fiat.u1(arg2), out1[2], arg1[2])
@@ -596,7 +597,8 @@ fe_set :: proc "contextless" (out1, arg1: ^Tight_Field_Element) {
 	out1[4] = x5
 }

-fe_cond_swap :: proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: int) {
+@(optimization_mode="none")
+fe_cond_swap :: #force_no_inline proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: int) {
 	mask := -u64(arg1)
 	x := (out1[0] ~ out2[0]) & mask
 	x1, y1 := out1[0] ~ x, out2[0] ~ x
@@ -201,7 +201,8 @@ fe_opp :: proc "contextless" (out1: ^Loose_Field_Element, arg1: ^Tight_Field_Ele
 	out1[2] = x3
 }

-fe_cond_assign :: proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: bool) {
+@(optimization_mode="none")
+fe_cond_assign :: #force_no_inline proc "contextless" (out1, arg1: ^Tight_Field_Element, arg2: bool) {
 	x1 := fiat.cmovznz_u64(fiat.u1(arg2), out1[0], arg1[0])
 	x2 := fiat.cmovznz_u64(fiat.u1(arg2), out1[1], arg1[1])
 	x3 := fiat.cmovznz_u64(fiat.u1(arg2), out1[2], arg1[2])
@@ -342,7 +343,8 @@ fe_set :: #force_inline proc "contextless" (out1, arg1: ^Tight_Field_Element) {
 	out1[2] = x3
 }

-fe_cond_swap :: proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: bool) {
+@(optimization_mode="none")
+fe_cond_swap :: #force_no_inline proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: bool) {
 	mask := -u64(arg1)
 	x := (out1[0] ~ out2[0]) & mask
 	x1, y1 := out1[0] ~ x, out2[0] ~ x
@@ -70,7 +70,7 @@ hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -149,7 +149,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -228,7 +228,7 @@ hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -307,7 +307,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -77,7 +77,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _blake2.update(&ctx, buf[:read])
        } 
@@ -77,7 +77,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _blake2.update(&ctx, buf[:read])
        } 
@@ -8,15 +8,23 @@ KEY_SIZE :: 32
 NONCE_SIZE :: 12
 XNONCE_SIZE :: 24

+@(private)
 _MAX_CTR_IETF :: 0xffffffff

+@(private)
 _BLOCK_SIZE :: 64
+@(private)
 _STATE_SIZE_U32 :: 16
+@(private)
 _ROUNDS :: 20

+@(private)
 _SIGMA_0 : u32 : 0x61707865
+@(private)
 _SIGMA_1 : u32 : 0x3320646e
+@(private)
 _SIGMA_2 : u32 : 0x79622d32
+@(private)
 _SIGMA_3 : u32 : 0x6b206574

 Context :: struct {
@@ -179,6 +187,7 @@ reset :: proc (ctx: ^Context) {
 	ctx._is_initialized = false
 }

+@(private)
 _do_blocks :: proc (ctx: ^Context, dst, src: []byte, nr_blocks: int) {
 	// Enforce the maximum consumed keystream per nonce.
 	//
@@ -441,6 +450,7 @@ _do_blocks :: proc (ctx: ^Context, dst, src: []byte, nr_blocks: int) {
 	}
 }

+@(private)
 _hchacha20 :: proc (dst, key, nonce: []byte) {
 	x0, x1, x2, x3 := _SIGMA_0, _SIGMA_1, _SIGMA_2, _SIGMA_3
 	x4 := util.U32_LE(key[0:4])
@@ -10,8 +10,10 @@ KEY_SIZE :: chacha20.KEY_SIZE
 NONCE_SIZE :: chacha20.NONCE_SIZE
 TAG_SIZE :: poly1305.TAG_SIZE

+@(private)
 _P_MAX :: 64 * 0xffffffff // 64 * (2^32-1)

+@(private)
 _validate_common_slice_sizes :: proc (tag, key, nonce, aad, text: []byte) {
 	if len(tag) != TAG_SIZE {
 		panic("crypto/chacha20poly1305: invalid destination tag size")
@@ -37,7 +39,10 @@ _validate_common_slice_sizes :: proc (tag, key, nonce, aad, text: []byte) {
 	}
 }

+@(private)
 _PAD: [16]byte
+
+@(private)
 _update_mac_pad16 :: #force_inline proc (ctx: ^poly1305.Context, x_len: int) {
 	if pad_len := 16 - (x_len & (16-1)); pad_len != 16 {
 		poly1305.update(ctx, _PAD[:pad_len])
@@ -26,6 +26,7 @@ compare_constant_time :: proc "contextless" (a, b: []byte) -> int {
 //
 // The execution time of this routine is constant regardless of the
 // contents of the memory being compared.
+@(optimization_mode="none")
 compare_byte_ptrs_constant_time :: proc "contextless" (a, b: ^byte, n: int) -> int {
 	x := mem.slice_ptr(a, n)
 	y := mem.slice_ptr(b, n)
@@ -65,7 +65,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -70,7 +70,7 @@ hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -149,7 +149,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -228,7 +228,7 @@ hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -307,7 +307,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -79,7 +79,7 @@ hash_stream_128_3 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -164,7 +164,7 @@ hash_stream_128_4 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -249,7 +249,7 @@ hash_stream_128_5 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -334,7 +334,7 @@ hash_stream_160_3 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -419,7 +419,7 @@ hash_stream_160_4 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -504,7 +504,7 @@ hash_stream_160_5 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -589,7 +589,7 @@ hash_stream_192_3 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -674,7 +674,7 @@ hash_stream_192_4 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -759,7 +759,7 @@ hash_stream_192_5 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -844,7 +844,7 @@ hash_stream_224_3 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -929,7 +929,7 @@ hash_stream_224_4 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -1014,7 +1014,7 @@ hash_stream_224_5 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -1099,7 +1099,7 @@ hash_stream_256_3 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -1184,7 +1184,7 @@ hash_stream_256_4 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -1270,7 +1270,7 @@ hash_stream_256_5 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        ctx.str_len = u32(len(buf[:read]))
        if read > 0 {
            update(&ctx, buf[:read])
@@ -70,7 +70,7 @@ hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -149,7 +149,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -228,7 +228,7 @@ hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -307,7 +307,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -77,7 +77,7 @@ hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -159,7 +159,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -241,7 +241,7 @@ hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -323,7 +323,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -64,7 +64,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
 	defer delete(buf)
 	read := 1
 	for read > 0 {
-	    read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
 	    if read > 0 {
 			update(&ctx, buf[:read])
 	    } 
@@ -68,7 +68,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -67,7 +67,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -8,6 +8,7 @@ import "core:mem"
 KEY_SIZE :: 32
 TAG_SIZE :: 16

+@(private)
 _BLOCK_SIZE :: 16

 sum :: proc (dst, msg, key: []byte) {
@@ -141,6 +142,7 @@ reset :: proc (ctx: ^Context) {
 	ctx._is_initialized = false
 }

+@(private)
 _blocks :: proc (ctx: ^Context, msg: []byte, final := false) {
 	n: field.Tight_Field_Element = ---
 	final_byte := byte(!final)
@@ -12,7 +12,7 @@ _rand_bytes :: proc (dst: []byte) {

 	for l > 0 {
 		to_read := min(l, _MAX_PER_CALL_BYTES)
-		ret := unix.sys_getrandom(raw_data(dst), to_read, 0)
+		ret := unix.sys_getrandom(raw_data(dst), uint(to_read), 0)
 		if ret < 0 {
 			switch os.Errno(-ret) {
 			case os.EINTR:
@@ -69,7 +69,7 @@ hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -145,7 +145,7 @@ hash_stream_160 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -221,7 +221,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -297,7 +297,7 @@ hash_stream_320 :: proc(s: io.Stream) -> ([DIGEST_SIZE_320]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -67,7 +67,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -74,7 +74,7 @@ hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -153,7 +153,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -232,7 +232,7 @@ hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -311,7 +311,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -73,7 +73,7 @@ hash_stream_224 :: proc(s: io.Stream) -> ([DIGEST_SIZE_224]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -152,7 +152,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -231,7 +231,7 @@ hash_stream_384 :: proc(s: io.Stream) -> ([DIGEST_SIZE_384]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -310,7 +310,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -73,7 +73,7 @@ hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -155,7 +155,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _sha3.update(&ctx, buf[:read])
        } 
@@ -233,7 +233,7 @@ init :: proc(ctx: ^Context, key: []byte, c_rounds, d_rounds: int) {
 }

 update :: proc(ctx: ^Context, data: []byte) {
-    assert(ctx.is_initialized, "crypto/siphash: Context is not initalized")
+    assert(ctx.is_initialized, "crypto/siphash: Context is not initialized")
    ctx.last_block = len(data) / 8 * 8
    ctx.buf = data
    i := 0
@@ -66,7 +66,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -70,7 +70,7 @@ hash_stream_256 :: proc(s: io.Stream) -> ([DIGEST_SIZE_256]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -146,7 +146,7 @@ hash_stream_512 :: proc(s: io.Stream) -> ([DIGEST_SIZE_512]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            update(&ctx, buf[:read])
        } 
@@ -71,7 +71,7 @@ hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _tiger.update(&ctx, buf[:read])
        } 
@@ -150,7 +150,7 @@ hash_stream_160 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _tiger.update(&ctx, buf[:read])
        } 
@@ -229,7 +229,7 @@ hash_stream_192 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _tiger.update(&ctx, buf[:read])
        } 
@@ -71,7 +71,7 @@ hash_stream_128 :: proc(s: io.Stream) -> ([DIGEST_SIZE_128]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _tiger.update(&ctx, buf[:read])
        } 
@@ -150,7 +150,7 @@ hash_stream_160 :: proc(s: io.Stream) -> ([DIGEST_SIZE_160]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _tiger.update(&ctx, buf[:read])
        } 
@@ -229,7 +229,7 @@ hash_stream_192 :: proc(s: io.Stream) -> ([DIGEST_SIZE_192]byte, bool) {
    defer delete(buf)
    read := 1
    for read > 0 {
-        read, _ = s->impl_read(buf)
+        read, _ = io.read(s, buf)
        if read > 0 {
            _tiger.update(&ctx, buf[:read])
        } 
@@ -11,6 +11,8 @@ package util
 */

 import "core:mem"
+// Keep vet happy
+_ :: mem

 // @note(bp): this can replace the other two
 cast_slice :: #force_inline proc "contextless" ($D: typeid/[]$DE, src: $S/[]$SE) -> D {
@@ -66,7 +66,7 @@ hash_stream :: proc(s: io.Stream) -> ([DIGEST_SIZE]byte, bool) {
 	defer delete(buf)
 	read := 1
 	for read > 0 {
-	    read, _ = s->impl_read(buf)
+	    read, _ = io.read(s, buf)
 	    if read > 0 {
 			update(&ctx, buf[:read])
 	    } 
@@ -6,8 +6,10 @@ import "core:mem"
 SCALAR_SIZE :: 32
 POINT_SIZE :: 32

+@(private)
 _BASE_POINT: [32]byte = {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}

+@(private)
 _scalar_bit :: #force_inline proc "contextless" (s: ^[32]byte, i: int) -> u8 {
 	if i < 0 {
 		return 0
@@ -15,6 +17,7 @@ _scalar_bit :: #force_inline proc "contextless" (s: ^[32]byte, i: int) -> u8 {
 	return (s[i>>3] >> uint(i&7)) & 1
 }

+@(private)
 _scalarmult :: proc (out, scalar, point: ^[32]byte) {
 	// Montgomery pseduo-multiplication taken from Monocypher.

@@ -0,0 +1,221 @@
+package debug_pe
+
+PE_SIGNATURE_OFFSET_INDEX_POS :: 0x3c
+PE_SIGNATURE :: u32le(0x0000_4550) // "PE\x00\x00"
+PE_SIGNATURE_STRING :: "PE\x00\x00"
+
+OPTIONAL_HEADER_MAGIC :: enum u16le {
+	PE32      = 0x010b,
+	PE32_PLUS = 0x020b,
+}
+
+Optional_Header_Base :: struct #packed {
+	magic:                          OPTIONAL_HEADER_MAGIC,
+	major_linker_version:           u8,
+	minor_linker_version:           u8,
+	size_of_code:                   u32le,
+	size_of_initialized_data:       u32le,
+	size_of_uninitialized_data:     u32le,
+	address_of_entry_point:         u32le,
+	base_of_code:                   u32le,
+}
+
+File_Header :: struct #packed {
+	machine:                 IMAGE_FILE_MACHINE,
+	number_of_sections:      u16le,
+	time_date_stamp:         u32le,
+	pointer_to_symbol_table: u32le,
+	number_of_symbols:       u32le,
+	size_of_optional_header: u16le,
+	characteristics:         IMAGE_FILE_CHARACTERISTICS,
+}
+
+Data_Directory :: struct #packed {
+	virtual_address: u32le,
+	size:            u32le,
+}
+
+Optional_Header32 :: struct #packed {
+	using base: Optional_Header_Base,
+	base_of_data:                   u32le,
+	image_base:                     u32le,
+	section_alignment:              u32le,
+	file_alignment:                 u32le,
+	major_operating_system_version: u16le,
+	minor_operating_system_version: u16le,
+	major_image_version:            u16le,
+	minor_image_version:            u16le,
+	major_subsystem_version:        u16le,
+	minor_subsystem_version:        u16le,
+	win32_version_value:            u32le,
+	size_of_image:                  u32le,
+	size_of_headers:                u32le,
+	check_sum:                      u32le,
+	subsystem:                      IMAGE_SUBSYSTEM,
+	dll_characteristics:            IMAGE_DLLCHARACTERISTICS,
+	size_of_stack_reserve:          u32le,
+	size_of_stack_commit:           u32le,
+	size_of_heap_reserve:           u32le,
+	size_of_heap_commit:            u32le,
+	loader_flags:                   u32le,
+	number_of_rva_and_sizes:        u32le,
+	data_directory:                 [16]Data_Directory,
+}
+
+Optional_Header64 :: struct #packed {
+	using base: Optional_Header_Base,
+	image_base:                     u64le,
+	section_alignment:              u32le,
+	file_alignment:                 u32le,
+	major_operating_system_version: u16le,
+	minor_operating_system_version: u16le,
+	major_image_version:            u16le,
+	minor_image_version:            u16le,
+	major_subsystem_version:        u16le,
+	minor_subsystem_version:        u16le,
+	win32_version_value:            u32le,
+	size_of_image:                  u32le,
+	size_of_headers:                u32le,
+	check_sum:                      u32le,
+	subsystem:                      IMAGE_SUBSYSTEM,
+	dll_characteristics:            IMAGE_DLLCHARACTERISTICS,
+	size_of_stack_reserve:          u64le,
+	size_of_stack_commit:           u64le,
+	size_of_heap_reserve:           u64le,
+	size_of_heap_commit:            u64le,
+	loader_flags:                   u32le,
+	number_of_rva_and_sizes:        u32le,
+	data_directory:                 [16]Data_Directory,
+}
+
+// .debug section
+Debug_Directory_Entry :: struct {
+	characteristics:     u32le,
+	time_date_stamp:     u32le,
+	major_version:       u16le,
+	minor_version:       u16le,
+	type:                IMAGE_DEBUG_TYPE,
+	size_of_data:        u32le,
+	address_of_raw_data: u32le,
+	pointer_to_raw_data: u32le,
+}
+
+
+IMAGE_FILE_MACHINE :: enum u16le {
+	UNKNOWN     = 0x0,
+	AM33        = 0x1d3,
+	AMD64       = 0x8664,
+	ARM         = 0x1c0,
+	ARMNT       = 0x1c4,
+	ARM64       = 0xaa64,
+	EBC         = 0xebc,
+	I386        = 0x14c,
+	IA64        = 0x200,
+	LOONGARCH32 = 0x6232,
+	LOONGARCH64 = 0x6264,
+	M32R        = 0x9041,
+	MIPS16      = 0x266,
+	MIPSFPU     = 0x366,
+	MIPSFPU16   = 0x466,
+	POWERPC     = 0x1f0,
+	POWERPCFP   = 0x1f1,
+	R4000       = 0x166,
+	SH3         = 0x1a2,
+	SH3DSP      = 0x1a3,
+	SH4         = 0x1a6,
+	SH5         = 0x1a8,
+	THUMB       = 0x1c2,
+	WCEMIPSV2   = 0x169,
+}
+
+// IMAGE_DIRECTORY_ENTRY constants
+IMAGE_DIRECTORY_ENTRY :: enum u8 {
+	EXPORT         = 0,
+	IMPORT         = 1,
+	RESOURCE       = 2,
+	EXCEPTION      = 3,
+	SECURITY       = 4,
+	BASERELOC      = 5,
+	DEBUG          = 6,
+	ARCHITECTURE   = 7, // reserved
+	GLOBALPTR      = 8,
+	TLS            = 9,
+	LOAD_CONFIG    = 10,
+	BOUND_IMPORT   = 11,
+	IAT            = 12,
+	DELAY_IMPORT   = 13,
+	COM_DESCRIPTOR = 14, // DLR Runtime headers
+	_RESERVED      = 15,
+}
+#assert(len(IMAGE_DIRECTORY_ENTRY) == 16)
+
+
+IMAGE_FILE_CHARACTERISTICS :: distinct bit_set[IMAGE_FILE_CHARACTERISTIC; u16le]
+IMAGE_FILE_CHARACTERISTIC :: enum u16le {
+	RELOCS_STRIPPED         = 0,
+	EXECUTABLE_IMAGE        = 1,
+	LINE_NUMS_STRIPPED      = 2,
+	LOCAL_SYMS_STRIPPED     = 3,
+	AGGRESIVE_WS_TRIM       = 4,
+	LARGE_ADDRESS_AWARE     = 5,
+
+	BYTES_REVERSED_LO       = 7,
+	MACHINE_32BIT           = 8, // IMAGE_FILE_32BIT_MACHINE  originally
+	DEBUG_STRIPPED          = 9,
+	REMOVABLE_RUN_FROM_SWAP = 10,
+	NET_RUN_FROM_SWAP       = 11,
+	SYSTEM                  = 12,
+	DLL                     = 13,
+	UP_SYSTEM_ONLY          = 14,
+	BYTES_REVERSED_HI       = 15,
+}
+
+IMAGE_SUBSYSTEM :: enum u16le {
+	UNKNOWN                  = 0,
+	NATIVE                   = 1,
+	WINDOWS_GUI              = 2,
+	WINDOWS_CUI              = 3,
+	OS2_CUI                  = 5,
+	POSIX_CUI                = 7,
+	NATIVE_WINDOWS           = 8,
+	WINDOWS_CE_GUI           = 9,
+	EFI_APPLICATION          = 10,
+	EFI_BOOT_SERVICE_DRIVER  = 11,
+	EFI_RUNTIME_DRIVER       = 12,
+	EFI_ROM                  = 13,
+	XBOX                     = 14,
+	WINDOWS_BOOT_APPLICATION = 16,
+}
+
+IMAGE_DLLCHARACTERISTICS :: distinct bit_set[IMAGE_DLLCHARACTERISTIC; u16le]
+IMAGE_DLLCHARACTERISTIC :: enum u16le {
+	HIGH_ENTROPY_VA       = 5,
+	DYNAMIC_BASE          = 6,
+	FORCE_INTEGRITY       = 7,
+	NX_COMPAT             = 8,
+	NO_ISOLATION          = 9,
+	NO_SEH                = 10,
+	NO_BIND               = 11,
+	APPCONTAINER          = 12,
+	WDM_DRIVER            = 13,
+	GUARD_CF              = 14,
+	TERMINAL_SERVER_AWARE = 15,
+}
+
+IMAGE_DEBUG_TYPE :: enum u32le {
+	UNKNOWN               = 0,  // An unknown value that is ignored by all tools.
+	COFF                  = 1,  // The COFF debug information (line numbers, symbol table, and string table). This type of debug information is also pointed to by fields in the file headers.
+	CODEVIEW              = 2,  // The Visual C++ debug information.
+	FPO                   = 3,  // The frame pointer omission (FPO) information. This information tells the debugger how to interpret nonstandard stack frames, which use the EBP register for a purpose other than as a frame pointer.
+	MISC                  = 4,  // The location of DBG file.
+	EXCEPTION             = 5,  // A copy of .pdata section.
+	FIXUP                 = 6,  // Reserved.
+	OMAP_TO_SRC           = 7,  // The mapping from an RVA in image to an RVA in source image.
+	OMAP_FROM_SRC         = 8,  // The mapping from an RVA in source image to an RVA in image.
+	BORLAND               = 9,  // Reserved for Borland.
+	RESERVED10            = 10, // Reserved.
+	CLSID                 = 11, // Reserved.
+	REPRO                 = 16, // PE determinism or reproducibility.
+	EX_DLLCHARACTERISTICS = 20, // Extended DLL characteristics bits.
+}
+
@@ -0,0 +1,128 @@
+package debug_pe
+
+Section_Header32 :: struct {
+	name:                    [8]u8,
+	virtual_size:            u32le,
+	virtual_address:         u32le,
+	size_of_raw_data:        u32le,
+	pointer_to_raw_data:     u32le,
+	pointer_to_relocations:  u32le,
+	pointer_to_line_numbers: u32le,
+	number_of_relocations:   u16le,
+	number_of_line_numbers:  u16le,
+	characteristics:         IMAGE_SCN_CHARACTERISTICS,
+}
+
+Reloc :: struct {
+	virtual_address:    u32le,
+	symbol_table_index: u32le,
+	type:               IMAGE_REL,
+}
+
+IMAGE_SCN_CHARACTERISTICS :: enum u32le {
+	TYPE_NO_PAD            = 0x00000008, // The section should not be padded to the next boundary. This flag is obsolete and is replaced by IMAGE_SCN_ALIGN_1BYTES. This is valid only for object files. = 0x00000010, // Reserved for future use.
+	CNT_CODE               = 0x00000020, // The section contains executable code.
+	CNT_INITIALIZED_DATA   = 0x00000040, // The section contains initialized data.
+	CNT_UNINITIALIZED_DATA = 0x00000080, // The section contains uninitialized data.
+	LNK_OTHER              = 0x00000100, // Reserved for future use.
+	LNK_INFO               = 0x00000200, // The section contains comments or other information. The .drectve section has this type. This is valid for object files only. = 0x00000400, // Reserved for future use.
+	LNK_REMOVE             = 0x00000800, // The section will not become part of the image. This is valid only for object files.
+	LNK_COMDAT             = 0x00001000, // The section contains COMDAT data. For more information, see COMDAT Sections (Object Only). This is valid only for object files.
+	GPREL                  = 0x00008000, // The section contains data referenced through the global pointer (GP).
+	MEM_PURGEABLE          = 0x00020000, // Reserved for future use.
+	MEM_16BIT              = 0x00020000, // Reserved for future use.
+	MEM_LOCKED             = 0x00040000, // Reserved for future use.
+	MEM_PRELOAD            = 0x00080000, // Reserved for future use.
+	ALIGN_1BYTES           = 0x00100000, // Align data on a 1-byte boundary. Valid only for object files.
+	ALIGN_2BYTES           = 0x00200000, // Align data on a 2-byte boundary. Valid only for object files.
+	ALIGN_4BYTES           = 0x00300000, // Align data on a 4-byte boundary. Valid only for object files.
+	ALIGN_8BYTES           = 0x00400000, // Align data on an 8-byte boundary. Valid only for object files.
+	ALIGN_16BYTES          = 0x00500000, // Align data on a 16-byte boundary. Valid only for object files.
+	ALIGN_32BYTES          = 0x00600000, // Align data on a 32-byte boundary. Valid only for object files.
+	ALIGN_64BYTES          = 0x00700000, // Align data on a 64-byte boundary. Valid only for object files.
+	ALIGN_128BYTES         = 0x00800000, // Align data on a 128-byte boundary. Valid only for object files.
+	ALIGN_256BYTES         = 0x00900000, // Align data on a 256-byte boundary. Valid only for object files.
+	ALIGN_512BYTES         = 0x00A00000, // Align data on a 512-byte boundary. Valid only for object files.
+	ALIGN_1024BYTES        = 0x00B00000, // Align data on a 1024-byte boundary. Valid only for object files.
+	ALIGN_2048BYTES        = 0x00C00000, // Align data on a 2048-byte boundary. Valid only for object files.
+	ALIGN_4096BYTES        = 0x00D00000, // Align data on a 4096-byte boundary. Valid only for object files.
+	ALIGN_8192BYTES        = 0x00E00000, // Align data on an 8192-byte boundary. Valid only for object files.
+	LNK_NRELOC_OVFL        = 0x01000000, // The section contains extended relocations.
+	MEM_DISCARDABLE        = 0x02000000, // The section can be discarded as needed.
+	MEM_NOT_CACHED         = 0x04000000, // The section cannot be cached.
+	MEM_NOT_PAGED          = 0x08000000, // The section is not pageable.
+	MEM_SHARED             = 0x10000000, // The section can be shared in memory.
+	MEM_EXECUTE            = 0x20000000, // The section can be executed as code.
+	MEM_READ               = 0x40000000, // The section can be read.
+	MEM_WRITE              = 0x80000000, // The section can be written to.
+}
+
+
+IMAGE_REL :: enum u16le {
+	I386_ABSOLUTE         = 0x0000,
+	I386_DIR16            = 0x0001,
+	I386_REL16            = 0x0002,
+	I386_DIR32            = 0x0006,
+	I386_DIR32NB          = 0x0007,
+	I386_SEG12            = 0x0009,
+	I386_SECTION          = 0x000A,
+	I386_SECREL           = 0x000B,
+	I386_TOKEN            = 0x000C,
+	I386_SECREL7          = 0x000D,
+	I386_REL32            = 0x0014,
+
+	AMD64_ABSOLUTE        = 0x0000,
+	AMD64_ADDR64          = 0x0001,
+	AMD64_ADDR32          = 0x0002,
+	AMD64_ADDR32NB        = 0x0003,
+	AMD64_REL32           = 0x0004,
+	AMD64_REL32_1         = 0x0005,
+	AMD64_REL32_2         = 0x0006,
+	AMD64_REL32_3         = 0x0007,
+	AMD64_REL32_4         = 0x0008,
+	AMD64_REL32_5         = 0x0009,
+	AMD64_SECTION         = 0x000A,
+	AMD64_SECREL          = 0x000B,
+	AMD64_SECREL7         = 0x000C,
+	AMD64_TOKEN           = 0x000D,
+	AMD64_SREL32          = 0x000E,
+	AMD64_PAIR            = 0x000F,
+	AMD64_SSPAN32         = 0x0010,
+
+	ARM_ABSOLUTE          = 0x0000,
+	ARM_ADDR32            = 0x0001,
+	ARM_ADDR32NB          = 0x0002,
+	ARM_BRANCH24          = 0x0003,
+	ARM_BRANCH11          = 0x0004,
+	ARM_SECTION           = 0x000E,
+	ARM_SECREL            = 0x000F,
+	ARM_MOV32             = 0x0010,
+
+	THUMB_MOV32           = 0x0011,
+	THUMB_BRANCH20        = 0x0012,
+	THUMB_BRANCH24        = 0x0014,
+	THUMB_BLX23           = 0x0015,
+
+	ARM_PAIR              = 0x0016,
+
+	ARM64_ABSOLUTE        = 0x0000,
+	ARM64_ADDR32          = 0x0001,
+	ARM64_ADDR32NB        = 0x0002,
+	ARM64_BRANCH26        = 0x0003,
+	ARM64_PAGEBASE_REL21  = 0x0004,
+	ARM64_REL21           = 0x0005,
+	ARM64_PAGEOFFSET_12A  = 0x0006,
+	ARM64_PAGEOFFSET_12L  = 0x0007,
+	ARM64_SECREL          = 0x0008,
+	ARM64_SECREL_LOW12A   = 0x0009,
+	ARM64_SECREL_HIGH12A  = 0x000A,
+	ARM64_SECREL_LOW12L   = 0x000B,
+	ARM64_TOKEN           = 0x000C,
+	ARM64_SECTION         = 0x000D,
+	ARM64_ADDR64          = 0x000E,
+	ARM64_BRANCH19        = 0x000F,
+	ARM64_BRANCH14        = 0x0010,
+	ARM64_REL32           = 0x0011,
+}
+
+PE_CODE_VIEW_SIGNATURE_RSDS :: u32le(0x5344_5352)
@@ -0,0 +1,108 @@
+package debug_pe
+
+COFF_SYMBOL_SIZE :: 18
+
+COFF_Symbol :: struct {
+	name:                  [8]u8,
+	value:                 u32le,
+	section_number:        i16le,
+	type:                  IMAGE_SYM_TYPE,
+	storage_class:         IMAGE_SYM_CLASS,
+	number_of_aux_symbols: u8,
+}
+
+// COFF_Symbol_Aux_Format5 describes the expected form of an aux symbol
+// attached to a section definition symbol. The PE format defines a
+// number of different aux symbol formats: format 1 for function
+// definitions, format 2 for .be and .ef symbols, and so on. Format 5
+// holds extra info associated with a section definition, including
+// number of relocations + line numbers, as well as COMDAT info. See
+// https://docs.microsoft.com/en-us/windows/win32/debug/pe-format#auxiliary-format-5-section-definitions
+// for more on what's going on here.
+COFF_Symbol_Aux_Format5 :: struct {
+	size:             u32le,
+	num_relocs:       u16le,
+	num_line_numbers: u16le,
+	checksum:         u32le,
+	sec_num:          u16le,
+	selection:        IMAGE_COMDAT_SELECT,
+	_:                [3]u8, // padding
+}
+
+IMAGE_COMDAT_SELECT :: enum u8 {
+	NODUPLICATES = 1,
+	ANY          = 2,
+	SAME_SIZE    = 3,
+	EXACT_MATCH  = 4,
+	ASSOCIATIVE  = 5,
+	LARGEST      = 6,
+}
+
+
+// The symbol record is not yet assigned a section. A value of zero indicates
+// that a reference to an external symbol is defined elsewhere. A value of
+// non-zero is a common symbol with a size that is specified by the value.
+IMAGE_SYM_UNDEFINED              :: 0
+// The symbol has an absolute (non-relocatable) value and is not an address.
+IMAGE_SYM_ABSOLUTE               :: -1
+// The symbol provides general type or debugging information but does not
+// correspond to a section. Microsoft tools use this setting along
+// with .file records (storage class FILE).
+IMAGE_SYM_DEBUG                  :: -2
+
+IMAGE_SYM_TYPE :: enum u16le {
+	NULL   = 0,
+	VOID   = 1,
+	CHAR   = 2,
+	SHORT  = 3,
+	INT    = 4,
+	LONG   = 5,
+	FLOAT  = 6,
+	DOUBLE = 7,
+	STRUCT = 8,
+	UNION  = 9,
+	ENUM   = 10,
+	MOE    = 11,
+	BYTE   = 12,
+	WORD   = 13,
+	UINT   = 14,
+	DWORD  = 15,
+	PCODE  = 32768,
+
+	DTYPE_NULL     = 0,
+	DTYPE_POINTER  = 0x10,
+	DTYPE_FUNCTION = 0x20,
+	DTYPE_ARRAY    = 0x30,
+}
+
+IMAGE_SYM_CLASS :: enum u8 {
+	NULL             = 0,
+	AUTOMATIC        = 1,
+	EXTERNAL         = 2,
+	STATIC           = 3,
+	REGISTER         = 4,
+	EXTERNAL_DEF     = 5,
+	LABEL            = 6,
+	UNDEFINED_LABEL  = 7,
+	MEMBER_OF_STRUCT = 8,
+	ARGUMENT         = 9,
+	STRUCT_TAG       = 10,
+	MEMBER_OF_UNION  = 11,
+	UNION_TAG        = 12,
+	TYPE_DEFINITION  = 13,
+	UNDEFINED_STATIC = 14,
+	ENUM_TAG         = 15,
+	MEMBER_OF_ENUM   = 16,
+	REGISTER_PARAM   = 17,
+	BIT_FIELD        = 18,
+	FAR_EXTERNAL     = 68, // Not in PECOFF v8 spec
+	BLOCK            = 100,
+	FUNCTION         = 101,
+	END_OF_STRUCT    = 102,
+	FILE             = 103,
+	SECTION          = 104,
+	WEAK_EXTERNAL    = 105,
+	CLR_TOKEN        = 107,
+
+	END_OF_FUNCTION  = 255,
+}
@@ -0,0 +1,7 @@
+/*
+Package core:dynlib implements loading of shared libraries/DLLs and their symbols.
+
+The behaviour of dynamically loaded libraries is specific to the target platform of the program.
+For in depth detail on the underlying behaviour please refer to your target platform's documentation.
+*/
+package dynlib
@@ -1,15 +1,94 @@
 package dynlib

+/*
+A handle to a dynamically loaded library.
+*/
 Library :: distinct rawptr

-load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
+/*
+Loads a dynamic library from the filesystem. The paramater `global_symbols` makes the symbols in the loaded
+library available to resolve references in subsequently loaded libraries.
+
+The paramater `global_symbols` is only used for the platforms `linux`, `darwin`, `freebsd` and `openbsd`.
+On `windows` this paramater is ignored.
+
+The underlying behaviour is platform specific.  
+On `linux`, `darwin`, `freebsd` and `openbsd` refer to `dlopen`.  
+On `windows` refer to `LoadLibraryW`.
+
+**Implicit Allocators**  
+`context.temp_allocator`
+
+Example:
+	import "core:dynlib"
+	import "core:fmt"
+
+	load_my_library :: proc() {
+		LIBRARY_PATH :: "my_library.dll"
+		library, ok := dynlib.load_library(LIBRARY_PATH)
+		if ! ok {
+			return
+		}
+		fmt.println("The library %q was successfully loaded", LIBRARY_PATH)
+	}
+*/
+load_library :: proc(path: string, global_symbols := false) -> (library: Library, did_load: bool) {
 	return _load_library(path, global_symbols)
 }

-unload_library :: proc(library: Library) -> bool {
+/*
+Unloads a dynamic library.
+
+The underlying behaviour is platform specific.  
+On `linux`, `darwin`, `freebsd` and `openbsd` refer to `dlclose`.  
+On `windows` refer to `FreeLibrary`.
+
+Example:
+	import "core:dynlib"
+	import "core:fmt"
+
+	load_then_unload_my_library :: proc() {
+		LIBRARY_PATH :: "my_library.dll"
+		library, ok := dynlib.load_library(LIBRARY_PATH)
+		if ! ok {
+			return
+		}
+		did_unload := dynlib.unload_library(library)
+		if ! did_unload {
+			return
+		}
+		fmt.println("The library %q was successfully unloaded", LIBRARY_PATH)
+	}
+*/
+unload_library :: proc(library: Library) -> (did_unload: bool) {
 	return _unload_library(library)
 }

+/*
+Loads the address of a procedure/variable from a dynamic library.
+
+The underlying behaviour is platform specific.  
+On `linux`, `darwin`, `freebsd` and `openbsd` refer to `dlsym`.  
+On `windows` refer to `GetProcAddress`.
+
+**Implicit Allocators**  
+`context.temp_allocator`
+
+Example:
+	import "core:dynlib"
+	import "core:fmt"
+
+	find_a_in_my_library :: proc() {
+		LIBRARY_PATH :: "my_library.dll"
+		library, ok := dynlib.load_library(LIBRARY_PATH)
+		if ! ok {
+			return
+		}
+
+		a, found_a := dynlib.symbol_address(library, "a")
+		if found_a do fmt.printf("The symbol %q was found at the address %v", "a", a)
+	}
+*/
 symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) #optional_ok {
 	return _symbol_address(library, symbol)
 }
@@ -0,0 +1,15 @@
+//+build js
+//+private
+package dynlib
+
+_load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
+	return nil, false
+}
+
+_unload_library :: proc(library: Library) -> bool {
+	return false
+}
+
+_symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) {
+	return nil, false
+}
@@ -4,10 +4,12 @@ package dynlib

 import win32 "core:sys/windows"
 import "core:strings"
+import "core:runtime"

 _load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
 	// NOTE(bill): 'global_symbols' is here only for consistency with POSIX which has RTLD_GLOBAL

+	runtime.DEFAULT_TEMP_ALLOCATOR_TEMP_GUARD()
 	wide_path := win32.utf8_to_wstring(path, context.temp_allocator)
 	handle := cast(Library)win32.LoadLibraryW(wide_path)
 	return handle, handle != nil
@@ -19,6 +21,7 @@ _unload_library :: proc(library: Library) -> bool {
 }

 _symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) {
+	runtime.DEFAULT_TEMP_ALLOCATOR_TEMP_GUARD()
 	c_str := strings.clone_to_cstring(symbol, context.temp_allocator)
 	ptr = win32.GetProcAddress(cast(win32.HMODULE)library, c_str)
 	found = ptr != nil
@@ -248,7 +248,7 @@ _read_record :: proc(r: ^Reader, dst: ^[dynamic]string, allocator := context.all
 					field_length += 1

 				case '\n', '\r':
-					if !is_quoted { break read_loop }
+					is_quoted or_break read_loop

 				case r.comma:
 					field_length = 0
@@ -42,7 +42,7 @@ write :: proc(w: ^Writer, record: []string) -> io.Error {
 				}
 			}
 		case:
-			if strings.contains_rune(field, w.comma) >= 0 {
+			if strings.contains_rune(field, w.comma) {
 				return true
 			}
 			if strings.contains_any(field, CHAR_SET) {
@@ -184,28 +184,26 @@ decode_xml :: proc(input: string, options := XML_Decode_Options{}, allocator :=

 advance :: proc(t: ^Tokenizer) -> (err: Error) {
 	if t == nil { return .Tokenizer_Is_Nil }
-	using t
-
 	#no_bounds_check {
-		if read_offset < len(src) {
-			offset = read_offset
-			r, w   = rune(src[read_offset]), 1
+		if t.read_offset < len(t.src) {
+			t.offset = t.read_offset
+			t.r, t.w   = rune(t.src[t.read_offset]), 1
 			switch {
-			case r == 0:
+			case t.r == 0:
 				return .Illegal_NUL_Character
-			case r >= utf8.RUNE_SELF:
-				r, w = utf8.decode_rune_in_string(src[read_offset:])
-				if r == utf8.RUNE_ERROR && w == 1 {
+			case t.r >= utf8.RUNE_SELF:
+				t.r, t.w = utf8.decode_rune_in_string(t.src[t.read_offset:])
+				if t.r == utf8.RUNE_ERROR && t.w == 1 {
 					return .Illegal_UTF_Encoding
-				} else if r == utf8.RUNE_BOM && offset > 0 {
+				} else if t.r == utf8.RUNE_BOM && t.offset > 0 {
 					return .Illegal_BOM
 				}
 			}
-			read_offset += w
+			t.read_offset += t.w
 			return .None
 		} else {
-			offset = len(src)
-			r = -1
+			t.offset = len(t.src)
+			t.r = -1
 			return
 		}
 	}
@@ -273,26 +271,25 @@ _extract_xml_entity :: proc(t: ^Tokenizer) -> (entity: string, err: Error) {
 		All of these would be in the ASCII range.
 		Even if one is not, it doesn't matter. All characters we need to compare to extract are.
 	*/
-	using t

 	length := len(t.src)
 	found  := false

 	#no_bounds_check {
-		for read_offset < length {
-			if src[read_offset] == ';' {
+		for t.read_offset < length {
+			if t.src[t.read_offset] == ';' {
+				t.read_offset += 1
 				found = true
-				read_offset += 1
 				break
 			}
-			read_offset += 1
+			t.read_offset += 1
 		}
 	}

 	if found {
-		return string(src[offset + 1 : read_offset - 1]), .None
+		return string(t.src[t.offset + 1 : t.read_offset - 1]), .None
 	}
-	return string(src[offset : read_offset]), .Invalid_Entity_Encoding
+	return string(t.src[t.offset : t.read_offset]), .Invalid_Entity_Encoding
 }

 /*
@@ -0,0 +1,73 @@
+package hex
+
+import "core:strings"
+
+encode :: proc(src: []byte, allocator := context.allocator) -> []byte #no_bounds_check {
+	dst := make([]byte, len(src) * 2, allocator)
+	for i, j := 0, 0; i < len(src); i += 1 {
+		v := src[i]
+		dst[j]   = HEXTABLE[v>>4]
+		dst[j+1] = HEXTABLE[v&0x0f]
+		j += 2
+	}
+
+	return dst
+}
+
+
+decode :: proc(src: []byte, allocator := context.allocator) -> (dst: []byte, ok: bool) #no_bounds_check {
+	if len(src) % 2 == 1 {
+		return
+	}
+
+	dst = make([]byte, len(src) / 2, allocator)
+	for i, j := 0, 1; j < len(src); j += 2 {
+		p := src[j-1]
+		q := src[j]
+
+		a := hex_digit(p) or_return
+		b := hex_digit(q) or_return
+
+		dst[i] = (a << 4) | b
+		i += 1
+	}
+
+	return dst, true
+}
+
+// Decodes the given sequence into one byte.
+// Should be called with one byte worth of the source, eg: 0x23 -> '#'.
+decode_sequence :: proc(str: string) -> (res: byte, ok: bool) {
+	str := str
+	if strings.has_prefix(str, "0x") || strings.has_prefix(str, "0X") {
+		str = str[2:]
+	}
+
+	if len(str) != 2 {
+		return 0, false
+	}
+
+	upper := hex_digit(str[0]) or_return
+	lower := hex_digit(str[1]) or_return
+
+	return upper << 4 | lower, true
+}
+
+@(private)
+HEXTABLE := [16]byte {
+	'0', '1', '2', '3',
+	'4', '5', '6', '7',
+	'8', '9', 'a', 'b',
+	'c', 'd', 'e', 'f',
+}
+
+@(private)
+hex_digit :: proc(char: byte) -> (u8, bool) {
+	switch char {
+	case '0' ..= '9': return char - '0', true
+	case 'a' ..= 'f': return char - 'a' + 10, true
+	case 'A' ..= 'F': return char - 'A' + 10, true
+	case:             return 0, false
+	}
+}
+
@@ -107,7 +107,7 @@ Node :: struct {
 /* Conventions */
 /* ------------
 Much of HxA's use is based on convention. HxA lets users store arbitrary data in its structure that can be parsed but whose semantic meaning does not need to be understood.
-A few conventions are hard, and some are soft. Hard convention that a user HAS to follow in order to produce a valid file. Hard conventions simplify parsing becaus the parser can make some assumptions. Soft convenbtions are basicly recomendations of how to store common data.
+A few conventions are hard, and some are soft. Hard convention that a user HAS to follow in order to produce a valid file. Hard conventions simplify parsing becaus the parser can make some assumptions. Soft convenbtions are basically recomendations of how to store common data.
 If you use HxA for something not covered by the conventions but need a convention for your use case. Please let us know so that we can add it!
 */

@@ -83,7 +83,7 @@ read :: proc(data: []byte, filename := "<input>", print_error := false, allocato
 		meta_data = make([]Meta, int(capacity))
 		count := 0
 		defer meta_data = meta_data[:count]
-		for m in &meta_data {
+		for &m in meta_data {
 			m.name = read_name(r) or_return

 			type := read_value(r, Meta_Value_Type) or_return
@@ -116,7 +116,7 @@ read :: proc(data: []byte, filename := "<input>", print_error := false, allocato
 		layer_count := 0
 		layers = make(Layer_Stack, stack_count)
 		defer layers = layers[:layer_count]
-		for layer in &layers {
+		for &layer in layers {
 			layer.name = read_name(r) or_return
 			layer.components = read_value(r, u8) or_return
 			type := read_value(r, Layer_Data_Type) or_return
@@ -5,6 +5,7 @@ import "core:math/bits"
 import "core:runtime"
 import "core:strconv"
 import "core:strings"
+import "core:reflect"
 import "core:io"

 Marshal_Data_Error :: enum {
@@ -17,25 +18,54 @@ Marshal_Error :: union #shared_nil {
 	io.Error,
 }

-marshal :: proc(v: any, allocator := context.allocator) -> (data: []byte, err: Marshal_Error) {
+// careful with MJSON maps & non quotes usage as keys without whitespace will lead to bad results
+Marshal_Options :: struct {
+	// output based on spec
+	spec: Specification,
+
+	// use line breaks & tab|spaces
+	pretty: bool, 
+
+	// spacing
+	use_spaces: bool,
+	spaces: int,
+
+	// state
+	indentation: int,
+
+	// option to output uint in JSON5 & MJSON
+	write_uint_as_hex: bool, 
+
+	// mjson output options
+	mjson_keys_use_quotes: bool,
+	mjson_keys_use_equal_sign: bool,
+
+	// mjson state
+	mjson_skipped_first_braces_start: bool,
+	mjson_skipped_first_braces_end: bool,
+}
+
+marshal :: proc(v: any, opt: Marshal_Options = {}, allocator := context.allocator) -> (data: []byte, err: Marshal_Error) {
 	b := strings.builder_make(allocator)
 	defer if err != nil {
 		strings.builder_destroy(&b)
 	}

-	marshal_to_builder(&b, v) or_return
+	opt := opt
+	marshal_to_builder(&b, v, &opt) or_return
 	
 	if len(b.buf) != 0 {
 		data = b.buf[:]
 	}
+
 	return data, nil
 }

-marshal_to_builder :: proc(b: ^strings.Builder, v: any) -> Marshal_Error {
-	return marshal_to_writer(strings.to_writer(b), v)
+marshal_to_builder :: proc(b: ^strings.Builder, v: any, opt: ^Marshal_Options) -> Marshal_Error {
+	return marshal_to_writer(strings.to_writer(b), v, opt)
 }

-marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
+marshal_to_writer :: proc(w: io.Writer, v: any, opt: ^Marshal_Options) -> (err: Marshal_Error) {
 	if v == nil {
 		io.write_string(w, "null") or_return
 		return
@@ -56,6 +86,7 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
 		case i16:     u = u128(i)
 		case i32:     u = u128(i)
 		case i64:     u = u128(i)
+		case i128:    u = u128(i)
 		case int:     u = u128(i)
 		case u8:      u = u128(i)
 		case u16:     u = u128(i)
@@ -82,7 +113,21 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
 		case u128be: u = u128(i)
 		}

-		s := strconv.append_bits_128(buf[:], u, 10, info.signed, 8*ti.size, "0123456789", nil)
+		s: string
+
+		// allow uints to be printed as hex
+		if opt.write_uint_as_hex && (opt.spec == .JSON5 || opt.spec == .MJSON) {
+			switch i in a {
+			case u8, u16, u32, u64, u128: 
+				s = strconv.append_bits_128(buf[:], u, 16, info.signed, 8*ti.size, "0123456789abcdef", { .Prefix })
+
+			case:
+				s = strconv.append_bits_128(buf[:], u, 10, info.signed, 8*ti.size, "0123456789", nil)
+			}
+		} else {
+			s = strconv.append_bits_128(buf[:], u, 10, info.signed, 8*ti.size, "0123456789", nil)
+		}
+
 		io.write_string(w, s) or_return


@@ -108,7 +153,7 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
 		case complex128: r, i = f64(real(z)), f64(imag(z))
 		case: return .Unsupported_Type
 		}
-	
+
 		io.write_byte(w, '[')    or_return
 		io.write_f64(w, r)       or_return
 		io.write_string(w, ", ") or_return
@@ -120,8 +165,8 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {

 	case runtime.Type_Info_String:
 		switch s in a {
-		case string:  io.write_quoted_string(w, s)         or_return
-		case cstring: io.write_quoted_string(w, string(s)) or_return
+		case string:  io.write_quoted_string(w, s, '"', nil, true)         or_return
+		case cstring: io.write_quoted_string(w, string(s), '"', nil, true) or_return
 		}

 	case runtime.Type_Info_Boolean:
@@ -147,10 +192,13 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
 	case runtime.Type_Info_Multi_Pointer:
 		return .Unsupported_Type

+	case runtime.Type_Info_Soa_Pointer:
+		return .Unsupported_Type
+
 	case runtime.Type_Info_Procedure:
 		return .Unsupported_Type

-	case runtime.Type_Info_Tuple:
+	case runtime.Type_Info_Parameters:
 		return .Unsupported_Type

 	case runtime.Type_Info_Simd_Vector:
@@ -159,95 +207,114 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
 	case runtime.Type_Info_Relative_Pointer:
 		return .Unsupported_Type

-	case runtime.Type_Info_Relative_Slice:
+	case runtime.Type_Info_Relative_Multi_Pointer:
 		return .Unsupported_Type
 		
 	case runtime.Type_Info_Matrix:
 		return .Unsupported_Type

 	case runtime.Type_Info_Array:
-		io.write_byte(w, '[') or_return
+		opt_write_start(w, opt, '[') or_return
 		for i in 0..<info.count {
-			if i > 0 { io.write_string(w, ", ") or_return }
-
+			opt_write_iteration(w, opt, i) or_return
 			data := uintptr(v.data) + uintptr(i*info.elem_size)
-			marshal_to_writer(w, any{rawptr(data), info.elem.id}) or_return
+			marshal_to_writer(w, any{rawptr(data), info.elem.id}, opt) or_return
 		}
-		io.write_byte(w, ']') or_return
+		opt_write_end(w, opt, ']') or_return
 		
 	case runtime.Type_Info_Enumerated_Array:
 		index := runtime.type_info_base(info.index).variant.(runtime.Type_Info_Enum)
-		io.write_byte(w, '[') or_return
+		opt_write_start(w, opt, '[') or_return
 		for i in 0..<info.count {
-			if i > 0 { io.write_string(w, ", ") or_return }
-
+			opt_write_iteration(w, opt, i) or_return
 			data := uintptr(v.data) + uintptr(i*info.elem_size)
-			marshal_to_writer(w, any{rawptr(data), info.elem.id}) or_return
+			marshal_to_writer(w, any{rawptr(data), info.elem.id}, opt) or_return
 		}
-		io.write_byte(w, ']') or_return
+		opt_write_end(w, opt, ']') or_return
 		
 	case runtime.Type_Info_Dynamic_Array:
-		io.write_byte(w, '[') or_return
+		opt_write_start(w, opt, '[') or_return
 		array := cast(^mem.Raw_Dynamic_Array)v.data
 		for i in 0..<array.len {
-			if i > 0 { io.write_string(w, ", ") or_return }
-
+			opt_write_iteration(w, opt, i) or_return
 			data := uintptr(array.data) + uintptr(i*info.elem_size)
-			marshal_to_writer(w, any{rawptr(data), info.elem.id}) or_return
+			marshal_to_writer(w, any{rawptr(data), info.elem.id}, opt) or_return
 		}
-		io.write_byte(w, ']') or_return
+		opt_write_end(w, opt, ']') or_return

 	case runtime.Type_Info_Slice:
-		io.write_byte(w, '[') or_return
+		opt_write_start(w, opt, '[') or_return
 		slice := cast(^mem.Raw_Slice)v.data
 		for i in 0..<slice.len {
-			if i > 0 { io.write_string(w, ", ") or_return }
-
+			opt_write_iteration(w, opt, i) or_return
 			data := uintptr(slice.data) + uintptr(i*info.elem_size)
-			marshal_to_writer(w, any{rawptr(data), info.elem.id}) or_return
+			marshal_to_writer(w, any{rawptr(data), info.elem.id}, opt) or_return
 		}
-		io.write_byte(w, ']') or_return
+		opt_write_end(w, opt, ']') or_return

 	case runtime.Type_Info_Map:
 		m := (^mem.Raw_Map)(v.data)
+		opt_write_start(w, opt, '{') or_return

-		io.write_byte(w, '{') or_return
 		if m != nil {
-			if info.generated_struct == nil {
+			if info.map_info == nil {
 				return .Unsupported_Type
 			}
-			entries    := &m.entries
-			gs         := runtime.type_info_base(info.generated_struct).variant.(runtime.Type_Info_Struct)
-			ed         := runtime.type_info_base(gs.types[1]).variant.(runtime.Type_Info_Dynamic_Array)
-			entry_type := ed.elem.variant.(runtime.Type_Info_Struct)
-			entry_size := ed.elem_size
+			map_cap := uintptr(runtime.map_cap(m^))
+			ks, vs, hs, _, _ := runtime.map_kvh_data_dynamic(m^, info.map_info)

-			for i in 0..<entries.len {
-				if i > 0 { io.write_string(w, ", ") or_return }
+			i := 0
+			for bucket_index in 0..<map_cap {
+				runtime.map_hash_is_valid(hs[bucket_index]) or_continue

-				data := uintptr(entries.data) + uintptr(i*entry_size)
-				key   := rawptr(data + entry_type.offsets[2])
-				value := rawptr(data + entry_type.offsets[3])
+				opt_write_iteration(w, opt, i) or_return
+				i += 1

-				marshal_to_writer(w, any{key, info.key.id})     or_return
-				io.write_string(w, ": ")                        or_return
-				marshal_to_writer(w, any{value, info.value.id}) or_return
+				key   := rawptr(runtime.map_cell_index_dynamic(ks, info.map_info.ks, bucket_index))
+				value := rawptr(runtime.map_cell_index_dynamic(vs, info.map_info.vs, bucket_index))
+
+				// check for string type
+				{
+					v := any{key, info.key.id}
+					ti := runtime.type_info_base(type_info_of(v.id))
+					a := any{v.data, ti.id}
+					name: string
+
+					#partial switch info in ti.variant {
+					case runtime.Type_Info_String:
+						switch s in a {
+						case string: name = s
+						case cstring: name = string(s)
+						}
+						opt_write_key(w, opt, name) or_return
+
+					case: return .Unsupported_Type
+					}
+				}
+
+				marshal_to_writer(w, any{value, info.value.id}, opt) or_return
 			}
 		}
-		io.write_byte(w, '}') or_return
+
+		opt_write_end(w, opt, '}') or_return

 	case runtime.Type_Info_Struct:
-		io.write_byte(w, '{') or_return
+		opt_write_start(w, opt, '{') or_return
+		
 		for name, i in info.names {
-			if i > 0 { io.write_string(w, ", ") or_return }
-			io.write_quoted_string(w, name) or_return
-			io.write_string(w, ": ") or_return
+			opt_write_iteration(w, opt, i) or_return
+			if json_name := string(reflect.struct_tag_get(auto_cast info.tags[i], "json")); json_name != "" {
+				opt_write_key(w, opt, json_name) or_return
+			} else {
+				opt_write_key(w, opt, name) or_return
+			}

 			id := info.types[i].id
 			data := rawptr(uintptr(v.data) + info.offsets[i])
-			marshal_to_writer(w, any{data, id}) or_return
+			marshal_to_writer(w, any{data, id}, opt) or_return
 		}
-		io.write_byte(w, '}') or_return
+
+		opt_write_end(w, opt, '}') or_return

 	case runtime.Type_Info_Union:
 		tag_ptr := uintptr(v.data) + info.tag_offset
@@ -270,11 +337,11 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {
 			io.write_string(w, "null") or_return
 		} else {
 			id := info.variants[tag-1].id
-			return marshal_to_writer(w, any{v.data, id})
+			return marshal_to_writer(w, any{v.data, id}, opt)
 		}

 	case runtime.Type_Info_Enum:
-		return marshal_to_writer(w, any{v.data, info.base.id})
+		return marshal_to_writer(w, any{v.data, info.base.id}, opt)

 	case runtime.Type_Info_Bit_Set:
 		is_bit_set_different_endian_to_platform :: proc(ti: ^runtime.Type_Info) -> bool {
@@ -330,3 +397,116 @@ marshal_to_writer :: proc(w: io.Writer, v: any) -> (err: Marshal_Error) {

 	return
 }
+
+// write key as quoted string or with optional quotes in mjson
+opt_write_key :: proc(w: io.Writer, opt: ^Marshal_Options, name: string) -> (err: io.Error)  {
+	switch opt.spec {
+	case .JSON, .JSON5:
+		io.write_quoted_string(w, name) or_return
+		io.write_string(w, ": " if opt.pretty else ":") or_return
+
+	case .MJSON:
+		if opt.mjson_keys_use_quotes {
+			io.write_quoted_string(w, name) or_return
+		} else {
+			io.write_string(w, name) or_return
+		}
+
+		if opt.mjson_keys_use_equal_sign {
+			io.write_string(w, " = " if opt.pretty else "=") or_return
+		} else {
+			io.write_string(w, ": " if opt.pretty else ":") or_return
+		}
+	}	
+
+	return
+}
+
+// insert start byte and increase indentation on pretty
+opt_write_start :: proc(w: io.Writer, opt: ^Marshal_Options, c: byte) -> (err: io.Error)  {
+	// skip mjson starting braces
+	if opt.spec == .MJSON && !opt.mjson_skipped_first_braces_start {
+		opt.mjson_skipped_first_braces_start = true
+		return
+	}
+
+	io.write_byte(w, c) or_return
+	opt.indentation += 1
+
+	if opt.pretty {
+		io.write_byte(w, '\n') or_return
+	}
+
+	return
+}
+
+// insert comma separation and write indentations
+opt_write_iteration :: proc(w: io.Writer, opt: ^Marshal_Options, iteration: int) -> (err: io.Error) {
+	switch opt.spec {
+	case .JSON, .JSON5: 
+		if iteration > 0 {
+			io.write_byte(w, ',') or_return
+
+			if opt.pretty {
+				io.write_byte(w, '\n') or_return
+			}
+		}
+
+		opt_write_indentation(w, opt) or_return
+
+	case .MJSON: 
+		if iteration > 0 {
+			// on pretty no commas necessary
+			if opt.pretty {
+				io.write_byte(w, '\n') or_return
+			} else {
+				// comma separation necessary for non pretty output!
+				io.write_byte(w, ',') or_return
+			}
+		}
+
+		opt_write_indentation(w, opt) or_return
+	}
+
+	return
+}
+
+// decrease indent, write spacing and insert end byte
+opt_write_end :: proc(w: io.Writer, opt: ^Marshal_Options, c: byte) -> (err: io.Error)  {
+	if opt.spec == .MJSON && opt.mjson_skipped_first_braces_start && !opt.mjson_skipped_first_braces_end {
+		if opt.indentation == 0 {
+			opt.mjson_skipped_first_braces_end = true
+			return
+		}
+	}
+
+	opt.indentation -= 1
+
+	if opt.pretty {
+		io.write_byte(w, '\n') or_return
+		opt_write_indentation(w, opt) or_return
+	}
+
+	io.write_byte(w, c) or_return
+	return
+}
+
+// writes current indentation level based on options
+opt_write_indentation :: proc(w: io.Writer, opt: ^Marshal_Options) -> (err: io.Error) {
+	if !opt.pretty {
+		return
+	}
+
+	if opt.use_spaces {
+		spaces := opt.spaces == 0 ? 4 : opt.spaces
+		for _ in 0..<opt.indentation * spaces {
+			io.write_byte(w, ' ') or_return
+		}
+	} else {
+		for _ in 0..<opt.indentation {
+			io.write_byte(w, '\t') or_return
+		}
+	}
+
+	return
+}
@@ -2,6 +2,7 @@ package json

 import "core:mem"
 import "core:unicode/utf8"
+import "core:unicode/utf16"
 import "core:strconv"

 Parser :: struct {
@@ -342,7 +343,7 @@ unquote_string :: proc(token: Token, spec: Specification, allocator := context.a
 			i += 1
 			continue
 		}
-		r, w := utf8.decode_rune_in_string(s)
+		r, w := utf8.decode_rune_in_string(s[i:])
 		if r == utf8.RUNE_ERROR && w == 1 {
 			break
 		}
@@ -403,11 +404,19 @@ unquote_string :: proc(token: Token, spec: Specification, allocator := context.a
 				}
 				i += 6

+				// If this is a surrogate pair, decode as such by taking the next rune too.
+				if r >= utf8.SURROGATE_MIN && r <= utf8.SURROGATE_HIGH_MAX && len(s) > i + 2 && s[i:i+2] == "\\u" {
+					r2 := get_u4_rune(s[i:])
+					if r2 >= utf8.SURROGATE_LOW_MIN && r2 <= utf8.SURROGATE_MAX {
+						i += 6
+						r = utf16.decode_surrogate_pair(r, r2)
+					}
+				}
+
 				buf, buf_width := utf8.encode_rune(r)
 				copy(b[w:], buf[:buf_width])
 				w += buf_width

-
 			case '0':
 				if spec != .JSON {
 					b[w] = '\x00'
@@ -163,8 +163,9 @@ get_token :: proc(t: ^Tokenizer) -> (token: Token, err: Error) {

 	skip_alphanum :: proc(t: ^Tokenizer) {
 		for t.offset < len(t.data) {
-			switch next_rune(t) {
+			switch t.r {
 			case 'A'..='Z', 'a'..='z', '0'..='9', '_':
+				next_rune(t)
 				continue
 			}

@@ -87,7 +87,8 @@ Error :: enum {



-destroy_value :: proc(value: Value) {
+destroy_value :: proc(value: Value, allocator := context.allocator) {
+	context.allocator = allocator
 	#partial switch v in value {
 	case Object:
 		for key, elem in v {
@@ -103,5 +104,4 @@ destroy_value :: proc(value: Value) {
 	case String:
 		delete(v)
 	}
-}
-
+}
@@ -72,7 +72,7 @@ unmarshal_string :: proc(data: string, ptr: ^$T, spec := DEFAULT_SPECIFICATION,
@(private)
 assign_bool :: proc(val: any, b: bool) -> bool {
 	v := reflect.any_core(val)
-	switch dst in &v {
+	switch &dst in v {
 	case bool: dst = bool(b)
 	case b8:   dst = b8  (b)
 	case b16:  dst = b16 (b)
@@ -85,7 +85,7 @@ assign_bool :: proc(val: any, b: bool) -> bool {
@(private)
 assign_int :: proc(val: any, i: $T) -> bool {
 	v := reflect.any_core(val)
-	switch dst in &v {
+	switch &dst in v {
 	case i8:      dst = i8     (i)
 	case i16:     dst = i16    (i)
 	case i16le:   dst = i16le  (i)
@@ -122,7 +122,7 @@ assign_int :: proc(val: any, i: $T) -> bool {
@(private)
 assign_float :: proc(val: any, f: $T) -> bool {
 	v := reflect.any_core(val)
-	switch dst in &v {
+	switch &dst in v {
 	case f16:     dst = f16  (f)
 	case f16le:   dst = f16le(f)
 	case f16be:   dst = f16be(f)
@@ -150,7 +150,7 @@ assign_float :: proc(val: any, f: $T) -> bool {
@(private)
 unmarshal_string_token :: proc(p: ^Parser, val: any, str: string, ti: ^reflect.Type_Info) -> bool {
 	val := val
-	switch dst in &val {
+	switch &dst in val {
 	case string:
 		dst = str
 		return true
@@ -215,6 +215,12 @@ unmarshal_value :: proc(p: ^Parser, v: any) -> (err: Unmarshal_Error) {
 		}
 	}
 	
+	switch &dst in v {
+	// Handle json.Value as an unknown type
+	case Value:
+		dst = parse_value(p) or_return
+		return
+	}
 	
 	#partial switch token.kind {
 	case .Null:
@@ -346,6 +352,8 @@ unmarshal_object :: proc(p: ^Parser, v: any, end_token: Token_Kind) -> (err: Unm
 			
 			fields := reflect.struct_fields_zipped(ti.id)
 			
+			runtime.DEFAULT_TEMP_ALLOCATOR_TEMP_GUARD(ignore = context.temp_allocator == context.allocator)
+
 			field_used := make([]bool, len(fields), context.temp_allocator)
 			
 			use_field_idx := -1
@@ -380,13 +388,18 @@ unmarshal_object :: proc(p: ^Parser, v: any, end_token: Token_Kind) -> (err: Unm
 				field := any{field_ptr, type.id}
 				unmarshal_value(p, field) or_return
 					
+				if parse_comma(p) {
+					break struct_loop
+				}
+				continue struct_loop
+			} else {
+				// allows skipping unused struct fields
+				parse_value(p) or_return
 				if parse_comma(p) {
 					break struct_loop
 				}
 				continue struct_loop
 			}
-			
-			return Unsupported_Type_Error{v.id, p.curr_token}
 		}
 		
 	case reflect.Type_Info_Map:
@@ -394,12 +407,10 @@ unmarshal_object :: proc(p: ^Parser, v: any, end_token: Token_Kind) -> (err: Unm
 			return UNSUPPORTED_TYPE
 		}
 		raw_map := (^mem.Raw_Map)(v.data)
-		if raw_map.entries.allocator.procedure == nil {
-			raw_map.entries.allocator = p.allocator
+		if raw_map.allocator.procedure == nil {
+			raw_map.allocator = p.allocator
 		}
 		
-		header := runtime.__get_map_header_runtime(raw_map, t)
-		
 		elem_backing := bytes_make(t.value.size, t.value.align, p.allocator) or_return
 		defer delete(elem_backing, p.allocator)
 		
@@ -415,19 +426,16 @@ unmarshal_object :: proc(p: ^Parser, v: any, end_token: Token_Kind) -> (err: Unm
 				delete(key, p.allocator)
 				return err
 			}
-			
-			hash := runtime.Map_Hash {
-				hash = runtime.default_hasher_string(&key, 0),
-				key_ptr = &key,
-			}
-			
+
+			key_ptr := rawptr(&key)
+
 			key_cstr: cstring
 			if reflect.is_cstring(t.key) {
 				key_cstr = cstring(raw_data(key))
-				hash.key_ptr = &key_cstr
+				key_ptr = &key_cstr
 			}
 			
-			set_ptr := runtime.__dynamic_map_set(header, hash, map_backing_value.data)
+			set_ptr := runtime.__dynamic_map_set_without_hash(raw_map, t.map_info, key_ptr, map_backing_value.data)
 			if set_ptr == nil {
 				delete(key, p.allocator)
 			} 
@@ -19,43 +19,39 @@ import "core:fmt"
 */
 print :: proc(writer: io.Writer, doc: ^Document) -> (written: int, err: io.Error) {
 	if doc == nil { return }
-	using fmt
-
-	written += wprintf(writer, "[XML Prolog]\n")
+	written += fmt.wprintf(writer, "[XML Prolog]\n")

 	for attr in doc.prologue {
-		written += wprintf(writer, "\t%v: %v\n", attr.key, attr.val)
+		written += fmt.wprintf(writer, "\t%v: %v\n", attr.key, attr.val)
 	}

-	written += wprintf(writer, "[Encoding] %v\n", doc.encoding)
+	written += fmt.wprintf(writer, "[Encoding] %v\n", doc.encoding)

 	if len(doc.doctype.ident) > 0 {
-		written += wprintf(writer, "[DOCTYPE]  %v\n", doc.doctype.ident)
+		written += fmt.wprintf(writer, "[DOCTYPE]  %v\n", doc.doctype.ident)

 		if len(doc.doctype.rest) > 0 {
-		 	wprintf(writer, "\t%v\n", doc.doctype.rest)
+		 	fmt.wprintf(writer, "\t%v\n", doc.doctype.rest)
 		}
 	}

 	for comment in doc.comments {
-		written += wprintf(writer, "[Pre-root comment]  %v\n", comment)
+		written += fmt.wprintf(writer, "[Pre-root comment]  %v\n", comment)
 	}

 	if len(doc.elements) > 0 {
-	 	wprintln(writer, " --- ")
+	 	fmt.wprintln(writer, " --- ")
 	 	print_element(writer, doc, 0)
-	 	wprintln(writer, " --- ")
+	 	fmt.wprintln(writer, " --- ")
 	 }

 	return written, .None
 }

 print_element :: proc(writer: io.Writer, doc: ^Document, element_id: Element_ID, indent := 0) -> (written: int, err: io.Error) {
-	using fmt
-
 	tab :: proc(writer: io.Writer, indent: int) {
 		for _ in 0..=indent {
-			wprintf(writer, "\t")
+			fmt.wprintf(writer, "\t")
 		}
 	}

@@ -64,22 +60,24 @@ print_element :: proc(writer: io.Writer, doc: ^Document, element_id: Element_ID,
 	element := doc.elements[element_id]

 	if element.kind == .Element {
-		wprintf(writer, "<%v>\n", element.ident)
-		if len(element.value) > 0 {
-			tab(writer, indent + 1)
-			wprintf(writer, "[Value] %v\n", element.value)
+		fmt.wprintf(writer, "<%v>\n", element.ident)
+
+		for value in element.value {
+			switch v in value {
+			case string:
+				tab(writer, indent + 1)
+				fmt.wprintf(writer, "[Value] %v\n", v)
+			case Element_ID:
+				print_element(writer, doc, v, indent + 1)
+			}
 		}

 		for attr in element.attribs {
 			tab(writer, indent + 1)
-			wprintf(writer, "[Attr] %v: %v\n", attr.key, attr.val)
-		}
-
-		for child in element.children {
-			print_element(writer, doc, child, indent + 1)
+			fmt.wprintf(writer, "[Attr] %v: %v\n", attr.key, attr.val)
 		}
 	} else if element.kind == .Comment {
-		wprintf(writer, "[COMMENT] %v\n", element.value)
+		fmt.wprintf(writer, "[COMMENT] %v\n", element.value)
 	}

 	return written, .None
--- a/Show More
+++ b/Show More