martong added a comment. In D74542#1874221 <https://reviews.llvm.org/D74542#1874221>, @teemperor wrote:
> In D74542#1874201 <https://reviews.llvm.org/D74542#1874201>, @martong wrote: > > > Looks good to me! Thanks! > > Adding @balazske as another reviewer though. He worked on something > > related to this, where we also experienced an assert in > > `isBeforeInTranslationUnit`. > > > IIRC the analyser is using the isBeforeInTranslationUnit for the error paths > so that's how the X-TU analysis could hit this situation (if the error path > is spanning Decls from different files). But that's mostly speculation. Yes exactly, but we actually use the SourceLocation from the "From" unit, we have a mapping in `CrossTranslationUnitContext` for that purpose. See `CrossTranslationUnitContext::getImportedFromSourceLocation` (and https://reviews.llvm.org/D64554 and https://reviews.llvm.org/D64638) Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D74542/new/ https://reviews.llvm.org/D74542 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits