danakj wrote: I am curious so I will ask: why did this take an approach of serializing the output of the pragmas (the maps) instead of serializing the pragmas themselves as AST nodes, so that the maps would be constructed in the same way as from parsing the text directly? that's how other pragmas were serialized that I have seen.
https://github.com/llvm/llvm-project/pull/92031 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
