mpark wrote: > > It turns out to be kind of a pain to use with HeaderFileInfoTrait. More > > importantly, we can't really afford to key by the internal key type since > > that only hashes on the size of the file and the hash collision gets pretty > > bad. Moreover, the merging and condensing strategy seems rather simple. > > Maybe we can use hash value of the input as the key?
By input, do you mean the file path? If so, as far as I understand we don't hash based that because different file paths will hash differently but they may actually be the same entity after all via symlinks. https://github.com/llvm/llvm-project/pull/140867 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
