[llvm] r211861 - ParseIR: don't take ownership of the MemoryBuffer

[Alp Toker]

On 27/06/2014 08:50, David Blaikie wrote:
> Give it a day or two, or a few weeks and no one will know why one
> MemoryBuffer is const and another isn't - it'll just be another
> inconsistency in the codebase without much in the way of a clue as to
> how/why it got there & whether or not it should be removed or kept
> around for some reason.

Okay, I've gone ahead and made the interface non-const as you suggest in 
LLVM r211883.

That means the callers now have to do the const_cast, which is covered 
by clang r211884.

If you have any more gardening ideas please do go ahead and commit directly.

Alp.

-- 
http://www.nuanti.com
the browser experts