[PATCH] Scoped NoAlias Metadata

[Hal Finkel]

----- Original Message -----
> From: "Chandler Carruth" <chandlerc at gmail.com>
> To: reviews+D2194+public+a9f2c71a9ad3b374 at llvm-reviews.chandlerc.com
> Cc: "Andy Trick" <atrick at apple.com>, "Dan Gohman" <sunfish at google.com>, "Chandler Carruth" <chandlerc at gmail.com>,
> "Hal Finkel" <hfinkel at anl.gov>, "Commit Messages and Patches for LLVM" <llvm-commits at cs.uiuc.edu>
> Sent: Saturday, January 11, 2014 2:11:57 AM
> Subject: Re: [PATCH] Scoped NoAlias Metadata
> 
> 
> 
> I'd still really like to hear Nick's thoughts about this based on the
> experience with lifetime markers...
> 

I spoke to Nick about this on IRC in January (Chandler was also involved in this conversation), the take-away from this was:

 1. No better design emerged.
 2. This will make it even more important that we have a sanitizer for aliasing (C99 restrict, etc).

Regarding (2), I completely agree, and I've committed to implementing this. I have an older implementation of this that I need to cleanup and submit. I'll do that in the near future.

Thanks again,
Hal

> 
> 
> 
> Anyways, on a totally unrelated topic:
> 
> 
> On Fri, Jan 10, 2014 at 11:12 PM, Andrew Trick < atrick at apple.com >
> wrote:
> 
> 
> I personally don't like indenting large class definitions within
> anonymous namespace. It's unnecessary indentation and I find myself
> wondering if I'm looking at a nested class.
> Yea, I think I'm just going to suggest a coding guideline to not
> indent namespaces. There has been a bit of back-and-forth on this,
> but I don't think there is nearly enough improved organization to
> justify the lost horizontal space.
> 
> 
> Helpfully, that is already clang-format's behavior. =]

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory