[PATCH] D135634: [clang][modules] Serialize VFS overlay paths into PCMs

Ben Barham via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Tue Oct 18 15:49:37 PDT 2022


bnbarham added a comment.

In D135634#3866704 <https://reviews.llvm.org/D135634#3866704>, @jansvoboda11 wrote:

> In D135634#3866353 <https://reviews.llvm.org/D135634#3866353>, @benlangmuir wrote:
>
>> I think we should deduplicate the vfs overlays if the same ivfsoverlay is specified in both the pcm and the command-line.
>
> My understanding is that `ASTUnit` never uses command-line and always adopts whatever is in the PCM.

It's now using the FS from `FileManager` as the base FS and I imagine was already just using the `FileManager` FS, so we are (and were) implicitly using the command-line overlays.

> In D135634#3866528 <https://reviews.llvm.org/D135634#3866528>, @bnbarham wrote:
>
>> Do reproducers read PCMs at all or do they rebuild from the headers?
>
> I don't know. How are they related to this patch?

It was more related to the chaining discussion. If reproducers do use PCMs then that would imply we *have* to chain, since we'd need to map from the path in the PCM -> original path (with PCM overlay) -> reproducer path (with reproducer overlay).

I suspect the answer is here is that they don't as it'd be easier to just include all the headers (I think?).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135634/new/

https://reviews.llvm.org/D135634



More information about the cfe-commits mailing list