On Tue, Dec 11, 2012 at 6:49 AM, Benyei, Guy <[email protected]> wrote: > Hi Eli, > Thanks for the review. > > Attached a fixed patch, please review. > >> + virtual bool hasFeature(StringRef Feature) const { >> + return Feature == "spir"; >> + } >> >> What's the point of this? > > This one is quite the same as in every virtual architecture, including > PNaCl/le32. I guess it is a convinient way to find out if the current target > implements this kind of abstraction.
Looks fine. (I was confused by your testcase for a little while, but then I figured out that the default optimization level is different for OpenCL compared to C.) -Eli _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
