fix: compiler warnings with no side effects

[bruno-dasilva]

Purpose

Fix the warning spam when compiling with clang, gcc, or msvc (mostly clang). This PR focuses on the non-behaviour changing fixes to code, mostly formatting changes or implicit casts.

AI Disclosure

This PR was generated through running claude code in a loop to fix every warning. However, every non-trivial change was inspected and several "fixes" were dropped in favor of different solutions to try and keep backwards compatibility in mind. I've called out the couple places where I don't understand the fix or the implication of the fix to other reviewers.

[bruno-dasilva]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• fix: memory barrier bitfield now pulls from correct parameter #2949
• fix: legit bugs found by compiler warnings #2946
• fix: compiler warnings with potential side effects #2934
• fix: compiler warnings with no side effects #2933 👈 (View in Graphite)
• fix: ignore compiler warnings for libs/submodules #2932
• fix: compiler warnings for missing overrides #2931
• master

This stack of pull requests is managed by Graphite. Learn more about stacking.

[sprunk]

what was the issue when these were missing?

[bruno-dasilva]

oh sorry, here's the warning (one per instantiated type):

  rts/System/type2.h:12:2: warning: instantiation of variable 'type2<int>::creg_class' required here,
          but no definition is available [-Wundefined-var-template]
     12 |         CR_DECLARE_STRUCT(type2)
        |         ^
  rts/System/creg/creg.h:244:33: note: expanded from macro 'CR_DECLARE_STRUCT'
  rts/System/creg/creg.h:216:46: note: expanded from macro 'CR_DECLARE_BASE'
    216 |         static creg::Class* StaticClass() { return &creg_class; } \
  probe.cpp:2:39: note: in instantiation of member function 'type2<int>::StaticClass' requested here
  rts/System/type2.h:12:2: note: forward declaration of template entity is here
  rts/System/creg/creg.h:213:21: note: expanded from macro 'CR_DECLARE_BASE'
    213 |         static creg::Class creg_class; \
  rts/System/type2.h:12:2: note: add an explicit instantiation declaration to suppress this warning
          if 'type2<int>::creg_class' is explicitly instantiated in another translation unit

just telling the compiler explicitly not to worry about generating these, as they're hand-written elsewhere. I think it's a limitation of compilation seeing only one file at a time, but then these resolve at linking time.

[sprunk]

Perhaps this should live in creg_cond.h then?

[bruno-dasilva]

Hm. Looking at cred_cond.h it's all type-agnostic creg plumbing, like its all macro definitions. Does it really make sense to put type2 stuff in there? Kinda inverts the dependency.

if you mean just factoring out the #ifdef USING_CREG we could write a helper macro in creg_cond.h that is something like #define CR_DECLARE_TEMPLATE_INSTANCE(TCls) template<> creg::Class TCls::creg_class

then in this type2 file you just use the macro:

CR_DECLARE_TEMPLATE_INSTANCE(int2)
CR_DECLARE_TEMPLATE_INSTANCE(float2)
...

wdyt?

[sprunk]

something like #define CR_DECLARE_TEMPLATE_INSTANCE(TCls) template<> creg::Class TCls::creg_class

Sounds alright.

[bruno-dasilva]

should be good now :)