kadircet added inline comments.
================
Comment at: clang-tools-extra/clangd/index/IndexAction.cpp:161
CI.getDiagnostics().hasUncompilableErrorOccurred()) {
llvm::errs() << "Skipping TU due to uncompilable errors\n";
+ } else {
----------------
sammccall wrote:
> I'm pretty sure the include graph is basically good in this case (up to the
> possibly-missing files). So we should at least send that, so invalidation
> works (or can work in the future).
>
> Also, we agreed not to overwrite the symbols for the included headers'
> shards, but what about the main file itself? Surely better to have something
> than nothing. (The "something" is roughly clang's recovery, which is what
> AST-based completions rely on anyway)
>
> More generally, do we need filtering logic at both ends?
> It seems like this file could just pass through the data it has, along with
> the info on whether there was an error (somehow)
Left the error detection behavior same(checking for uncompilable errors)
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D63986/new/
https://reviews.llvm.org/D63986
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
