On Wed, Apr 2, 2014 at 10:03 AM, Aaron Ballman <[email protected]>wrote:
> On Wed, Apr 2, 2014 at 12:28 PM, Nico Rieck <[email protected]> wrote: > > That's because this patch is Sema-only, and the dllimport function is > > still getting linkonce_odr linkage instead of available_externally. > > I can try to split out the relevant IRGen changes and tests and put them > > up for review sooner. > > That'd make me feel slightly more comfortable, if you can work it that way. > > The only other thing I'd like to see is a test case for an inline > function in a class (like my example from above). I didn't see > anything in our existing tests, but I may have simply overlooked it. > > Other than my minor nits, the patch LGTM, but I'd prefer if it was > committed after the IRGen changes are in. > I agree. This patch removes warnings about our ignorance of dllimport / dllexport inline functions, and we don't want to have a window where we emit incorrect IR without diagnostics. If you want to keep things incremental, you can probably remove the "attribute ignored" warnings when you fix the IRGen for dll-visible inline functions. Feel free to commit with that change if that works for you. Thanks for working on this!
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
