vsapsai added a comment.

The issue with this change is that it claims to add functionality that exists 
already. I.e., the test is passing without the change. The confusing part might 
be that even if `DirectoryLookup::LookupFile` doesn't find a file for relative 
destination immediately, it updates `MappedName` which is used by 
`HeaderSearch::LookupFile`. And I haven't tested it but looks like the change 
violates that header search paths are tested in order. So when we have `-I A -I 
B.hmap -I C` instead of processing A, B.hmap, C in that order, we can end up 
with A, B.hmap, A, B.hmap, C.

Unfortunately, https://crbug.com/1056174 doesn't describe a reproducible 
problem but only a proposed solution. If you provide more details about 
distributed compilation pitfalls, which paths are relative and which are 
absolute, we'll be able to help you with header maps.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D75323/new/

https://reviews.llvm.org/D75323



_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
  • [PATCH]... Reid "On Leave" Kleckner via Phabricator via cfe-commits
    • [P... Reid "On Leave" Kleckner via Phabricator via cfe-commits
    • [P... Volodymyr Sapsai via Phabricator via cfe-commits
    • [P... Nico Weber via Phabricator via cfe-commits
    • [P... Volodymyr Sapsai via Phabricator via cfe-commits

Reply via email to