arnetheduck planned changes to this revision.
arnetheduck added a comment.
Thanks for the comments - I'll see when I have time to get back to this as the
current patch happens to take care of my (rather limited) use case..
================
Comment at: lib/Format/UnwrappedLineParser.cpp:1086
@@ +1085,3 @@
+ int parens = 0;
+ while (!eof()) {
+ if (FormatTok->isOneOf(tok::l_paren, tok::l_square)) {
----------------
djasper wrote:
> We need to be much more error resilient here as people will frequently call
> clang-format with mismatched parentheses and such. I am actually quite
> skeptical about having yet another instance of parentheses counting. Maybe
> just call parseParens() instead?
makes sense.
================
Comment at: lib/Format/UnwrappedLineParser.cpp:1094
@@ +1093,3 @@
+ } else if (!FormatTok->isOneOf(tok::identifier, tok::coloncolon)) {
+ return false;
+ }
----------------
djasper wrote:
> This seems wrong. Even in the case of the bug you say you are fixing, the
> "23" is neither an identifier nor a coloncolon..
Ah, right... I've run it against our code base at work where it happened to be
enough.. , but since full expressions can be used nowadays, I guess it's a lot
more than just these to that *can* appear.
http://reviews.llvm.org/D18793
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
