Fznamznon wrote: > IIUC, we should fix these issue in the lexer phase? What do you think?
Well, I thought about this at first and I'm not an expert in neither Parser nor Lexer TBH. But to avoid possible regressions I need to be able to understand that I'm within a place where a kernel call is not possible, i.e. after operator keyword. I'm not sure I can ensure that from Lexer. Another argument for doing that in Parser is that we already do similar things in parser when it comes to template lists in https://github.com/llvm/llvm-project/blob/e43dcc851cbfacd38e2c2f13d7728925f19eafec/clang/lib/Parse/ParseTemplate.cpp#L1016 Please let me know if I'm not making any sense. > https://github.com/oneapi-src/SYCLomatic/blob/24bfedbdb9afbb951be47e3ab95b2abe5fbbe913/clang/lib/Lex/Lexer.cpp#L4276-L4296. Well that is the opposite of what I'm doing, this is merging tokens separated by space and I'm doing a split https://github.com/llvm/llvm-project/pull/218384 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
