Tracking issue for enabling `-Zdrop-tracking` by default

[eholk]

We tried to do this a while ago (#91032) but discovered a number of regressions soon afterwards so it was disabled (#93165). This issue tracks known regressions and their fixes. Note that many regressions have been fixed but are not listed here since they were done before I filed this bug.

• Drop tracking error message regression in issue-70935-complex-spans.rs #97332
• must_not_suspend lint regression with -Zdrop-tracking #97333
• drop-tracking: ICE when using default struct update syntax #98476
• drop-tracking: type parameters are incorrectly required to be Send to make futures Send #98477
• [drop tracking] Use parent expression for scope, not parent node  #101217
• Drop tracking ICE involving await in left hand side of assignment (-Zdrop-tracking) #102383
• DO NOT MERGE (yet): Enable -Zdrop-tracking by default #97334

[eholk]

@rustbot label +A-async-await +AsyncAwait-Triaged

[eholk]

@rustbot claim

[jyn514]

The last two issues have both been fixed :) and the first has been mitigated (it doesn't have the same error as it used to, but the new error is much more readable than previously).
Let me know if there's any way I can help with the must_not_suspend lint.

[eholk]

I suspect the first issue can be marked as fixed too. I'll double check that it still looks okay and if so go ahead and close the issue.

For must_not_suspend, if we're lucky this may just need someone to try it out, since it seems like #98754 might have fixed that as well.

[davidbarsky]

For must_not_suspend, if we're lucky this may just need someone to try it out, since it seems like #98754 might have fixed that as well.

Would it be helpful if tracing added #[must_not_suspend] behind an RUSTFLAGS? I can then try enabling it at work's monorepo and see what breaks. I'm not sure if we have #98754, but I can double-check.

[jyn514]

I think all of these have been fixed :) @eholk what are the next steps? Another Crater run?

[eholk]

@jyn514 - Yup! Let me rebase #97334 and then we should be able to run just the failed ones from last time.