kuilpd wrote: > I have a feeling that this would actually be better off as two node types > (float and integer), as each stage (lexer, parser and evaluator) handles the > two differently. Maybe then the integer node could use llvm::APInt directly, > which would make the `is_unsigned` field look less out of place. > > The implementation makes sense to me, but I don't know if we've reached > consensus that this is the way this should work.
The node will also have boolean literals and nullptr later, they are all stored in a `Scalar`, so I don't think we should do 4 nodes for this. I can add them in this PR as well. We didn't really discuss how literals are parsed and stored, only the math and type promotion part, which I will get to in the next PR. https://github.com/llvm/llvm-project/pull/152308 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
