aaboud added a comment. In https://reviews.llvm.org/D16135#669416, @aprantl wrote:
> In https://reviews.llvm.org/D16135#669045, @aaboud wrote: > > > Addressed Adrian last comments. > > Added a LIT tests that covers all the macro kinds: > > > > 1. built-in (define) > > 2. command-line (define, include, undef) > > 3. main source (define, include, undef) Checked the above with and without > > PCH include. > > > > Notice, that current implementation does not support debug info for macro > > definition and inclusion generated during the PCH file creation. To support > > that, we need to extend the PCH format to preserve this information. If you > > believe this feature is important, I will open a bugzilla ticket and we can > > handle it separately. > > > That would be a good idea and important to fix. I don't think we want the the > generated code (or the debug info for that matter) to change just because the > user decides to switch to PCH. Ideally turning on PCH should be transparent. > > I have one final question: What's the impact of this change on, e.g., > building clang? > > - How much does the build directory grow? > - Is there any noticeable compile time regression? I will run self-build and come back with answers for these questions. I am just wondering, if I build in debug mode, will it use the flag "-fstandalone-debug"? This flag is set by default only for DarwinClang and FreeBSD, is there a way to assure it will be used with self-build? Notice that without this flag, there will be no debug info macro generated. https://reviews.llvm.org/D16135 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
