Anastasia added a comment. In http://reviews.llvm.org/D17821#393496, @bader wrote:
> In http://reviews.llvm.org/D17821#393387, @Anastasia wrote: > > > Regarding, extending this approach for OpenCL pipe types too. I was > > thinking we could change current implementation to have ReadPipeType and > > WritePipeType. They can both be derived from PipeType that we already have > > now (we can make it an abstract class to avoid its instantiation?). > > > > Similarly to images, since read and write pipes will be mapped to different > > Clang types, we won't need any extra semantical checking but just need to > > add extra code in CodeGen of pipe type and builtins to accept two separate > > types for read only and write only cases. > > > > Would this make sense? > > > Sure. Do you want me to add it here or it's okay if we fix pipes in a > separate patch? Yes, I think it's better to go in a separate commit, not to complicate this one too much. Also since there are not many comment here, I think we should try to commit it ASAP otherwise rebasing would be an issue. http://reviews.llvm.org/D17821 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
