Hi, rom what I recall it was felt that even though these types are precisely defined in the OpenCL standard (unlike normal C) it was better that each OpenCL implementation should set them within its own subtarget at the same time it sets the data layout string (so they're definitely known to be consistent). Since doing this was code consolidation rather than new functionality we've gone with the consensus and abandoned the patch. (If consensus were to change I wouldn't object to a version without the address spaces going into clang.)
Cheers, Dave -----Original Message----- From: Pekka Jääskeläinen [mailto:[email protected]] Sent: 08 August 2013 10:34 To: [email protected] Cc: David Tweed; [email protected]; [email protected]; Chuck Zhao; Kalle Raiskila Subject: Re: [PATCH v3] Set some OpenCl specification mandated types/alignments/etc Ping? Here's a direct link to the patch. It still looks valid to me (except for the address space map part? *). http://lists.cs.uiuc.edu/pipermail/cfe-commits/attachments/20121127/7bb9509f /attachment.obj from http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20121126/068665.h tml (*) http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20121203/069460.h tml On 07/31/2013 12:54 AM, Pekka Jääskeläinen wrote: > Does someone know what happened to the patch that was going to fix the > widths of some of the OpenCL C data types? > > http://article.gmane.org/gmane.comp.compilers.clang.scm/61965/match=opencl+s p > ecification+mandated+types+alignments+etc -- Pekka _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
