 Ryan Fleury
|
981dd2e02c
|
tighten up usage of ctrl entity ctx vs. r/w ctx store vs. accel
|
2025-05-14 19:16:06 -07:00 |
|
|
Ryan Fleury
|
1b26c3f180
|
entity_ctx * entity_store -> entity_ctx, entity_ctx_rw_store, thread-local entity_ctx_lookup_accel
|
2025-05-14 19:11:02 -07:00 |
|
|
Ryan Fleury
|
1e6b3ee2e6
|
split read-only ctrl entity operations from read-write ctrl entity operations, represent in type system via EntityCtx vs. EntityStore
|
2025-05-14 18:35:25 -07:00 |
|
|
Ryan Fleury
|
e513af5361
|
ditto
|
2025-05-14 15:30:49 -07:00 |
|
|
Ryan Fleury
|
60e47fb821
|
unwind cache -> callstack cache
|
2025-05-14 15:28:57 -07:00 |
|
|
Ryan Fleury
|
81483f5100
|
eliminate per-callstack-frame 'parent num' - redundant info
|
2025-05-14 14:41:13 -07:00 |
|
|
Ryan Fleury
|
1e53ac29ef
|
still protect against infinite unwinds, but require that both rsp/rip remain unchanged, rather than just rip
|
2025-05-14 13:55:16 -07:00 |
|
|
Ryan Fleury
|
99a8108b2a
|
eliminate now-unnecessary debug info references in call stack data structure
|
2025-05-14 13:53:24 -07:00 |
|
|
Ryan Fleury
|
a7b6e6f02a
|
eliminate usage of old overcomplicated call stack data structure, which required actively holding debug info references... we want to eliminate this so that we can trivially cache call stacks / unwinds
|
2025-05-14 13:51:59 -07:00 |
|
|
Ryan Fleury
|
b49f4559ce
|
get main unwinding path off of requiring entire ctrl entity store
|
2025-05-14 13:32:32 -07:00 |
|
|
Ryan Fleury
|
f12b66c1ee
|
sketch out asynchronous unwinding stuff
|
2025-05-14 13:27:10 -07:00 |
|
|
Ryan Fleury
|
e60e0c8555
|
sketch out unwind cache
|
2025-05-14 11:43:54 -07:00 |
|
|
Ryan Fleury
|
a67f326f78
|
eliminate redundant recomputation of ir extensions
|
2025-05-14 10:27:01 -07:00 |
|
|
Ryan Fleury
|
10c4dcecd1
|
do not prohibit same-rips from stopping unwinds
|
2025-05-14 10:15:42 -07:00 |
|
|
Ryan Fleury
|
2e2eb4a6f2
|
bump to 18
|
2025-05-14 09:53:50 -07:00 |
|
|
Ryan Fleury
|
70ae35cbd7
|
do not use worker thread arenas from main thread in rdi baker...
|
2025-05-14 09:51:46 -07:00 |
|
|
Ryan Fleury
|
141b96c17e
|
do not skip 0 with texture formats
|
2025-05-14 08:21:54 -07:00 |
|
|
Ryan Fleury
|
a371374d6c
|
require file existence before calling into dwrite
|
2025-05-14 08:19:30 -07:00 |
|
|
Ryan Fleury
|
ad714e9baa
|
dwrite: do not try to create font face if font file creation fails
|
2025-05-14 08:11:13 -07:00 |
|
|
Ryan Fleury
|
271e8e4523
|
deduplicate new file path maps on-add
|
2025-05-13 16:29:23 -07:00 |
|
|
Ryan Fleury
|
5532aa6903
|
rdi_from_pdb: when relative paths are supplied by line info, build them as being relative to OBJ; this is not necessarily correct but it is the best thing we can do given the available information. the user needs to remap this once in the debugger, but then it will work.
|
2025-05-13 16:23:31 -07:00 |
|
|
Ryan Fleury
|
258b45a837
|
eliminate a bit of duplicate font cache lookups in code view build, + eliminate unnecessary per-box color lookups
|
2025-05-13 15:11:41 -07:00 |
|
|
Ryan Fleury
|
e10f0489da
|
record which irtree-and-type generations were produced via auto-hooks; skip those when doing typeofs
|
2025-05-13 14:19:30 -07:00 |
|
|
Ryan Fleury
|
89e93459d7
|
fix ufcs-style 'raw' lens application
|
2025-05-13 13:51:56 -07:00 |
|
|
Ryan Fleury
|
8b8c88f310
|
preserve whole chain of parent irtrees, do not simply apply only to autohook overrides; straighten out some event consumption weirdness in interaction between autocompletion floating view & query floating view; ensure we mark the cfg space as dirty when modifying strings
|
2025-05-13 13:44:14 -07:00 |
|
|
Ryan Fleury
|
97c58e3747
|
start sketching out pieces for flattened expansions in rows(...)
|
2025-05-13 11:55:06 -07:00 |
|
|
Ryan Fleury
|
34304280c9
|
fix old usage of table in mule_main; resolve to value in (sequence)
|
2025-05-13 10:55:39 -07:00 |
|
|
Ryan Fleury
|
11c1b5277e
|
only -> rows; table -> columns
|
2025-05-13 10:44:06 -07:00 |
|
|
Ryan Fleury
|
73c91ae2bc
|
eval: adjust array expansion, use parent key of pointer argument, rather than key of pointer argument, to scope count expression evaluation; fix up a few edge cases with $. and implicit $ member accesses; make omit much more powerful, by allowing list of arguments to control all expansion expressions, rather than assuming only member names
|
2025-05-13 10:19:47 -07:00 |
|
|
Ryan Fleury
|
86dcec4334
|
Merge branch 'dev' of https://github.com/EpicGames/raddebugger
|
2025-05-12 21:39:46 -07:00 |
|
|
Ryan Fleury
|
bc75635a65
|
os/gfx/linux: window focus querying
|
2025-05-12 21:22:03 -07:00 |
|
|
Ryan Fleury
|
4c1585444c
|
be less ambitious about occupying basic type names in debugger; can conflict with common purposes in codebases, and really we should defer to debug info when possible
|
2025-05-12 17:33:57 -07:00 |
|
|
Ryan Fleury
|
533136875e
|
eliminate dummy window egl setup
|
2025-05-12 17:20:46 -07:00 |
|
|
Ryan Fleury
|
d564184d31
|
chip away more unneeded path processing, impeding proper support for relative paths; we want to mostly preserve paths as we see them, and then we should find a different story for mapping the paths in very few places
|
2025-05-12 17:09:13 -07:00 |
|
|
Ryan Fleury
|
a5b227a1c6
|
egl/glx work; eliminate unneeded path normalization paths; do not assume os current path when normalizing paths; gl synchronous debug strings
|
2025-05-12 16:03:36 -07:00 |
|
|
Ryan Fleury
|
fe3cac7ac3
|
parameterize irtree generation path with identifier resolution rules; in most cases, we want the usual order: implicit accesses -> locals -> registers -> globals/tlocals/types/procedures -> macros; but if we are specifically evaluating a call expression tree, we want to prefer callables - in this case, macros should be prioritized.
|
2025-05-12 11:56:57 -07:00 |
|
|
Ryan Fleury
|
8688322a43
|
more of the same
|
2025-05-11 21:47:24 -07:00 |
|
|
Ryan Fleury
|
2307d44506
|
forward declare XLookupString ourselves, since it is apparently missing in libx11-dev
|
2025-05-11 21:26:39 -07:00 |
|
|
Ryan Fleury
|
cc319190c6
|
render/opengl: flushed instance buffers
|
2025-05-11 15:26:05 -07:00 |
|
|
Ryan Fleury
|
1896ec2a85
|
more os/gfx linux coverage
|
2025-05-11 15:19:40 -07:00 |
|
|
Ryan Fleury
|
684402efac
|
x11 event coverage
|
2025-05-11 14:14:12 -07:00 |
|
|
Ryan Fleury
|
9de203e2cd
|
adjust smooth-unhinted dwrite raster params
|
2025-05-11 13:48:05 -07:00 |
|
|
Ryan Fleury
|
8b3f07c7b5
|
eliminate font fudge factors; fix size units in freetype backend; fix directory iteration lack-of-robustness-to-failed-dir-open
|
2025-05-11 13:27:59 -07:00 |
|
|
Ryan Fleury
|
6df96b7eef
|
preserve mouse coordinates from xlib button events
|
2025-05-10 17:08:40 -07:00 |
|
|
Ryan Fleury
|
0a3869981f
|
correctly calculate shortened hash w/ dirty marker in build.sh
|
2025-05-10 16:01:56 -07:00 |
|
|
Ryan Fleury
|
219e5bd211
|
first pass at standing up freetype font provider backend
|
2025-05-10 15:59:44 -07:00 |
|
|
Ryan Fleury
|
868f2d0660
|
be a bit more robust to failures in window creation, IPC thread launching, & semaphore non-implemented parts
|
2025-05-10 14:39:32 -07:00 |
|
|
Ryan Fleury
|
3fd1f9d255
|
stub out demon/fontprovider so we can build debugger
|
2025-05-10 14:28:14 -07:00 |
|
|
Ryan Fleury
|
b52e3be5ae
|
fill out more linux os layer coverage
|
2025-05-10 11:38:43 -07:00 |
|
|
Ryan Fleury
|
b9350c867c
|
egl backend for linux backend of opengl backend of render
|
2025-05-10 11:04:59 -07:00 |
|