[PATCH] Use attachLowHighPC also for compile unit DIE

[Frédéric Riss]

> On 28 Aug 2014, at 20:01, David Blaikie <dblaikie at gmail.com> wrote:
> 
> 
> 
> 
> On Thu, Aug 28, 2014 at 9:42 AM, Frédéric Riss <friss at apple.com <mailto:friss at apple.com>> wrote:
> 
>> On 28 Aug 2014, at 17:08, David Blaikie <dblaikie at gmail.com <mailto:dblaikie at gmail.com>> wrote:
>> 
>> Could you add test coverage and attach the entire patch (covering all source and test changes together) as a single file for review?
>> 
> 
> Sure. I thought you would like the const patch split out.
> 
> Ah, sure - makes sense. You're welcome to commit that straight-up/now.

Will do thanks.

> Here’s the full diff with a little test.
> 
> Do we have existing test coverage for the subprogram case? I imagine we do somewhere (try breaking it and see which tests fail) though it might not be a nice/tidy as your test case. Perhaps you could add the CU coverage to the same test case (& possibly clean it up a bit if it's not as legible as your new case)

I can try to do that.

> Also, in cases where we include essentially untouched IR straight from the frontend, it's preferable to include the original C (or C++ or whatever) source that was used to produce the IR, makes it a bit easier to maintain if we then change the schema and need to update a test, understand what it was testing, etc. See other recently added debug info tests for examples of this.

Yeah, I know that and I did so in the few tests I worked on. This one was so trivial “void foo() {}” that I didn’t include it. But you’re right, it doesn’t cost anything to include it anyway (even more so when it’s so small).

Fred

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140828/fef799f2/attachment.html>