dexonsmith added a comment. In D104465#2825343 <https://reviews.llvm.org/D104465#2825343>, @dexonsmith wrote:
> A few other ideas: > > 1. Disable PCH validation when minimizing. (Is that less robust than your > current workaround? If so, can you explain why?) > 2. Use the original PCH header in the scanning `-cc1`s (translate > `-include-pch` to `-include`) and switch back in the generated `-cc1`s (back > to `-include-pch`). > 3. Embed instructions in the PCH for how to build it, and make a "minimized" > version of the PCH. Two more options to consider: 4. If a compilation uses a PCH, use the original files for any input file transitively referenced by the PCH. Can be done by using a filesystem overlay that skips over the minimization layer for those files, or could tell the minimization layer not to minimize those files. You can figure out the files by adding a free function that preloads the PCH and extracts out all the input files. 5. Add support for building/using a PCH, and only support PCHes that are built this way. E.g., add a `-include-pch-auto` option or something. Scanner would use `-include`, spit out a command for building a PCH using explicit modules, and the generated `-cc1`s would use `-include-pch` to refer to it. This way all the modules are generated "in house". Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D104465/new/ https://reviews.llvm.org/D104465 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
