keith added a comment. In D109128#2997588 <https://reviews.llvm.org/D109128#2997588>, @JDevlieghere wrote:
> If I understand correctly, this patch does that, but swaps 2 and 3. What's > the motivation for trying the non-canonical path first? If we treat the > current working directory as a property directory (which is what the > canonicalization is all about), then it seems like we should exhaust those > options first. Using the canonical path first is the root of the problem, using that first the behavior to return the absolute path virtually always, for example in my new test these assertions EXPECT_EQ("a", Name.get()); EXPECT_EQ("a", OpenedS->getName()); fail and would have to be converted to: EXPECT_EQ("//root/foo/a", Name.get()); EXPECT_EQ("//root/foo/a", OpenedS->getName()); Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D109128/new/ https://reviews.llvm.org/D109128 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits