transpile: Rewrite fn compare_src_locs implementation to have a total order

[Kriskras99]

This implementation is simplified compared to the previous one. It is also almost twice as slow in the exhaustive test (15 vs 25 seconds).
However, in real sort usage the impact should be significantly less.

• Fixes SrcLoc sorting is non-transitive and a false total order and equality (panics in 1.81) #1126

[kkysen]

Thanks! This looks a lot better, and the better comments help a lot as well.

Is there a test we can add to make sure this is working as intended (correctly sorted, no panics)?

[Kriskras99]

I've added a test for the total order, based on a reduced test case from the issue (apparently creduce can reduce any text file).
I've also included your code suggestions.

[Kriskras99]

Not sure what my IDE did to the imports but I'll fix it, I'll also fix your other suggestions. But it will have to wait until after the weekend

[kkysen]

Not sure what my IDE did to the imports but I'll fix it, I'll also fix your other suggestions. But it will have to wait until after the weekend

No problem, no rush.

[Kriskras99]

@kkysen I've included your suggestions. Is there anything else that needs to change?

[Kriskras99]

Added your last suggestion

[kkysen]

CI seems to be failing, only on Darwin, for some reason:

CI

/Users/runner/work/1/s/tests/statics/src/thread_locals.c:4:17: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
static __thread gsti = 37;
~~~~~~~~~~~~~~~ ^
int
/Users/runner/work/1/s/tests/statics/src/thread_locals.c:5:17: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
extern __thread geti;
~~~~~~~~~~~~~~~ ^
int
/Users/runner/work/1/s/tests/statics/src/thread_locals.c:12:21: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
    static __thread fsti = 59;
    ~~~~~~~~~~~~~~~ ^
    int
/Users/runner/work/1/s/tests/statics/src/thread_locals.c:13:21: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
    extern __thread feti;
    ~~~~~~~~~~~~~~~ ^
    int

This doesn't seem to be related to the changes, here, though, so I'm not sure what's happening.

[Kriskras99]

Maybe the darwin image got updated or something? I really don't see how this change can cause this

[Kriskras99]

Could you try triggering the CI on the last commit in master? See if it has the same failure

[Kriskras99]

@kkysen with your CI fix from yesterday, this PR is now green again.

[kkysen]

@kkysen with your CI fix from yesterday, this PR is now green again.

Awesome! I'll merge this now. Sorry for the delay with the CI, and thanks for fixing this!

[kkysen]

Oops, I'm now getting a different CI error in the CI workflow that won't run on outside contributors' PRs correctly: https://github.com/immunant/c2rust/actions/runs/11493361902/job/31988919249. Do you have an idea of what this might be? Sorry this CI workflow still isn't set up to correctly run on external PRs.

[Kriskras99]

Maybe a path confusion? If you look at the first error:

error[E0422]: cannot find struct, variant or union type `C2RustUnnamed_4` in this scope
    --> src/lib/conncache.rs:3264:38
     |
3264 |                 __sigaction_handler: C2RustUnnamed_4 {
     |                                      ^^^^^^^^^^^^^^^ not found in this scope
     |
help: consider importing one of these items
     |
3293 |     use crate::src::lib::conncache::C2RustUnnamed_4;
     |
       and 28 other candidates

It seems that C2RustUnnamed_4 is put in src/src/lib/conncache.rs instead of src/lib/conncache.rs and therefore requiring an import. But I haven't looked into c2rust-testsuite enough to understand what it's doing.