tom-anders added inline comments.
================
Comment at: clang-tools-extra/clangd/CodeComplete.cpp:1993
Builder.emplace(Recorder ? &Recorder->CCSema->getASTContext() :
nullptr,
- Item, SemaCCS, QueryScopes, *Inserter, FileName,
+ Recorder ? Recorder->CCSema->CurContext : nullptr,
Item,
+ SemaCCS, QueryScopes, *Inserter, FileName,
----------------
tom-anders wrote:
> kadircet wrote:
> > instead of passing it here, can we just do the traversal as part of
> > `getQueryScopes` and let these be handled uniformly by the shortest
> > qualifier logic ?
> Yeah that was my initial idea, however the problem is that it looses the
> information about whether the enum is scoped or not:
> 1) QueryScopes only stores strings
> 2) The index has no flag for indicating whether an EnumConstant belongs to a
> scoped or unscoped enum
>
> Thus, we get a small problem, consider the following code:
>
> ```
> namespace ns {
> enum class Scoped { Foo };
> enum class Unscoped { Bar };
> }
>
> int main() {
> using ns::Scoped;
> using ns::Unscoped;
>
> Foo^
> Bar^
> }
> ```
> Here we want to complete Foo^ to ns::Scoped::Foo, but for Bar^ we want only
> ns::Bar. To resolve this, the scoped/unscoped information needs to be
> available when building the completion candidate.
>
> Hope this was clear, maybe I missed something.
Sorry, I meant the completion for Foo^ should be Scoped::Foo here, not
ns::Scoped::Foo.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D141800/new/
https://reviews.llvm.org/D141800
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
