nikic added a comment.

In D139006#3988227 <https://reviews.llvm.org/D139006#3988227>, @sebastian-ne 
wrote:

> In D139006#3988215 <https://reviews.llvm.org/D139006#3988215>, @mkazantsev 
> wrote:
>
>> So now every single test needs to be regenerated? It'll create straw diff 
>> from nowhere...
>
> We don’t need to regenerate every test.
> Similar to how we don’t reformat all of LLVM after a change in clang-format, 
> we can just do that when regenerating tests when they have changes anyway.
> So, the churn is that we have extra changes when regenerating a test for a 
> patch.

This kind of change requires generating check lines in a separate commit prior 
to the patch, because functional patches must not contain this kind of noise. 
This kind of change to UTC output is very annoying.

Please revert this patch pending further discussion.

I think there's three options here:

1. The status quo, which is that you must pass `--function-signature` if you 
encounter this issue.
2. What this patch does, which causes massive churn.
3. Something like adding `--function-signature` to UTC_ARGS by default for new 
tests, while not doing so for old ones.

If we are not content with the status quo, then we should do something along 
the lines of 3. Possibly not with `--function-signature` but something like a 
`--version N` flag that defaults to the latest but gets persisted in UTC_ARGS. 
This way we are free to improve UTC without being worried about test churn.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D139006/new/

https://reviews.llvm.org/D139006

_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to