fix eval commit rules for strings; do not treat inputs as textual data when committing to pointers unless contents are quoted - fixes inability to change string pointer values

src/mule/mule_main.cpp

@@ -431,6 +431,7 @@ type_coverage_eval_tests(void)
     char data[4];
   };
   SomeDataStructured *some_data = (SomeDataStructured *)&some_data_with_a_string[0];
+  char *string_ptr = &string[0];
   
   const char *const_string = "Hello, World!";
   const char const_string_array[] = "Hello, World!";

src/raddbg/raddbg_core.c

@@ -1237,10 +1237,18 @@ rd_commit_eval_value_string(E_Eval dst_eval, String8 string)
         commit_data.size = Min(commit_data.size, e_type_byte_size_from_key(type_key));
       }
       
+      //- rjf: determine if commit string is quoted
+      B32 is_quoted = 0;
+      if(string.size >= 1 && string.str[0] == '"')
+      {
+        string = str8_skip(string, 1);
+        is_quoted = 1;
+      }
+      
       //- rjf: pointer or array to characters/integers? -> try to treat
       // new value string as textual data
       if(!got_commit_data &&
-         ((type_kind == E_TypeKind_Ptr || type_kind == E_TypeKind_Array) &&
+         (((is_quoted && type_kind == E_TypeKind_Ptr) || type_kind == E_TypeKind_Array) &&
           (direct_type_kind == E_TypeKind_Char8 ||
            direct_type_kind == E_TypeKind_Char16 ||
            direct_type_kind == E_TypeKind_Char32 ||
@@ -1250,12 +1258,6 @@ rd_commit_eval_value_string(E_Eval dst_eval, String8 string)
            e_type_kind_is_integer(direct_type_kind))))
       {
         got_commit_data = 1;
-        B32 is_quoted = 0;
-        if(string.size >= 1 && string.str[0] == '"')
-        {
-          string = str8_skip(string, 1);
-          is_quoted = 1;
-        }
         if(string.size >= 1 && string.str[string.size-1] == '"')
         {
           string = str8_chop(string, 1);
@@ -1306,9 +1308,9 @@ rd_commit_eval_value_string(E_Eval dst_eval, String8 string)
            src_eval_value.irtree.mode == E_Mode_Value)
         {
           got_commit_data = 1;
-          commit_data = push_str8_copy(scratch.arena, str8_struct(&src_eval.value));
+          commit_data = str8_copy(scratch.arena, str8_struct(&src_eval.value));
-          commit_data.size = Min(commit_data.size, e_type_byte_size_from_key(src_eval.irtree.type_key));
+          commit_data.size = Max(commit_data.size, e_type_byte_size_from_key(src_eval.irtree.type_key));
-          commit_data.size = Min(commit_data.size, e_type_byte_size_from_key(type_key));
+          commit_data.size = Max(commit_data.size, e_type_byte_size_from_key(type_key));
         }
       }
     }

src/raddbg/raddbg_main.c

@@ -5,21 +5,24 @@
 //~ rjf: post-0.9.20 TODO notes
 //
 //- fabian complaints / reports
-// [x] `foo, x` needs to correctly match `hex(foo)`, e.g. in application to expansions
 // [ ] broadly, need to complete memory view & fix issues
 //  [ ] dumb panel click-through thing
 //  [ ] correct type interpretations; togglable interpretations
 //  [ ] make address editable; specialized version of the cursor address editor in tab right-click menu
-// [x] eval: pointer casts of register space should promote to process space
 // [ ] simplified default layout
-// [ ] signify empty watch window "expression" slot more as a text field?
-// [ ] ambiguous parsing cases - need to re-introduce identifier resolution into eval parse. (foo *)&bar
-// [ ] more control over string visualization; specifically, when *not* to do it, even when using e.g. char *s
-// [ ] broadly, view discoverability / docs
 //  [ ] memory view deserves larger spot, in default layout
-// [ ] sign bits should not display on unsigned integers?
 // [ ] fix for-loop stepping oddities, likely single-line for-loop stepping
+// [x] `foo, x` needs to correctly match `hex(foo)`, e.g. in application to expansions
+// [x] eval: pointer casts of register space should promote to process space
+// [x] more control over string visualization; specifically, when *not* to do it, even when using e.g. char *s
+// [x] sign bits should not display on unsigned integers?
+// [ ] ambiguous parsing cases - need to re-introduce identifier resolution into eval parse. (foo *)&bar
+// [ ] broadly, view discoverability / docs
 // [ ] source/disasm ctx menu should include options for e.g. "go to selected thread"
+// [ ] signify empty watch window "expression" slot more as a text field?
+//
+//- jeff viz notes
+// [ ] single-line viz for pointers w/ bad (unmapped) addresses
 //
 //- namespace/locations/variables RDI pass
 // [ ] RDI_Local, RDI_GlobalVariable, RDI_ThreadVariable -> RDI_Variable