[PATCH] D67901: [clangd] Improve semantic highlighting in dependent contexts (fixes #154)

[Haojian Wu via Phabricator via cfe-commits]

hokein added inline comments.


================
Comment at: clang-tools-extra/clangd/SemanticHighlighting.cpp:219
+  bool VisitDependentNameTypeLoc(DependentNameTypeLoc L) {
+    addToken(L.getNameLoc(), HighlightingKind::DependentType);
+    return true;
----------------
nridge wrote:
> ilya-biryukov wrote:
> > nridge wrote:
> > > hokein wrote:
> > > > nit: we have `kindForType` for hanlding all types, so I'd move the logic of detecting the dependent type there.
> > > I did try this, but it doesn't quite work, because `VisitTypeLoc` adds the highlighting to the `TypeLoc`'s `getBeginLoc()`. For something like `typename T::type`, that highlights the `typename` token rather than the `type` token. By contrast, here I add the highlighting to the `DependentNameTypeLoc`'s `getNameLoc()` which will correctly highlight the `type` token.
> > You'd want to implement `WalkUpFromDependentNameTypeLoc` instead of `Visit*` to avoid adding extra highlightings in `VisitTypeLoc`.
> > 
> > In fact, I'm surprised we're not seeing them now.
> We're not seeing extra highlightings with the current patch, because `VisitTypeLoc` does not add any highlightings for dependent types (`kindForType()` returns `None` for them). 
> 
> So, I don't think there's a problem with using `VisitDependentNameTypeLoc`?
I'm second on Ilya's suggestion. It follows the existing pattern (see  WalkupFrom* methods above) and make the code clearer, and we can move the `DependentType` to `kindForType`.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D67901/new/

https://reviews.llvm.org/D67901