Athosvk added a comment. In https://reviews.llvm.org/D43667#1063049, @juliehockett wrote:
> In https://reviews.llvm.org/D43667#1062746, @Athosvk wrote: > > > I'm a bit late on this, but I'd say that YAML is usually not a 'final' > > format. What would be the use-cases for this? And if is meant as an > > alternative intermediate format, why not instead of having one big file, > > have one file per input file? Just wondering what the particular motivation > > for that could be > > > The idea is that it's a generally-consumable output format, and so could be > interpreted by external software fairly trivially. The bitsream format is > compact and good for things that will live entirely in the clang-doc tool, > but is harder to deal with outside that scope. YAML bridges that gap. That's what I expected :). The primary advantage of one file per output to us was granularity. That allows you to do incremental builds and distribute them at this stage (locally over multiple cores, but also remotely if need be). https://reviews.llvm.org/D43667 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
