Re: r279485 - Module debug info: Don't assert when encountering an incomplete definition

2016-08-23 Thread David Blaikie via cfe-commits
Might be good to mention in an email on this thread what the revision number of the missing test is (& hopefully that test commit also mentions the revision number of this change as well - so people can track through in either direction) On Tue, Aug 23, 2016 at 8:31 AM Adrian Prantl wrote: > Yes

Re: r279485 - Module debug info: Don't assert when encountering an incomplete definition

2016-08-23 Thread Adrian Prantl via cfe-commits
Yes, that's because I forgot to git-add the actual testcase :-) It came commits after this one. -- adrian > On Aug 22, 2016, at 11:00 PM, David Blaikie wrote: > > Generally I'd expect a test case to exercise/demonstrate the behavior that > was previously hidden behind an assertion. (put another

Re: r279485 - Module debug info: Don't assert when encountering an incomplete definition

2016-08-22 Thread David Blaikie via cfe-commits
Generally I'd expect a test case to exercise/demonstrate the behavior that was previously hidden behind an assertion. (put another way: "we expect something specific to happen, not just for clang not to assert/crash") On Mon, Aug 22, 2016 at 3:32 PM Adrian Prantl via cfe-commits < cfe-commits@list

r279485 - Module debug info: Don't assert when encountering an incomplete definition

2016-08-22 Thread Adrian Prantl via cfe-commits
Author: adrian Date: Mon Aug 22 17:23:58 2016 New Revision: 279485 URL: http://llvm.org/viewvc/llvm-project?rev=279485&view=rev Log: Module debug info: Don't assert when encountering an incomplete definition in isDefinedInClangModule() and assume that the incomplete definition is not defined in th