[llvm-commits] [PATCH] update getPointerTo to handle multiple address spaces

[Micah Villmow]

Ok, I'll see what I can do to deal with this. It most likely will require
merging this patch with the 001c-datalayout-cleanup.txt and other changes
to keep from breaking anything and adding the proper asserts everywhere.
On Oct 29, 2012 3:02 PM, "Chandler Carruth" <chandlerc at google.com> wrote:

> First let me correct a simple misunderstanding...
>
> On Mon, Oct 29, 2012 at 2:48 PM, Villmow, Micah <Micah.Villmow at amd.com>
> wrote:
> >> 1) The type used to "base" a pointer type on must *always* be a
> >> pointer. Please go back and implement my (and Duncan, and others')
> >> suggestion to assert if the type is anything other than a pointer or
> >> vector of pointers. This bit of code review has been neglected for
> >> over a week as churn and chaos develop in the codebase. That is
> >> unacceptable.
> > This was not true about the code base before I started work, so why
> would it be different now?
>
> I think you're reading this wrong (and frankly, it's just confusing to
> read...)
>
> Given the code "Ty->getPointerTo(OtherTy)", I expect to get a pointer
> to Ty, and use OtherTy to compute the address space. I think "OtherTy"
> should *have* to be a pointer if used in this way. Clearly, "Ty" can
> be anything, it simply ends up as the element type of the pointer.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20121029/092c2084/attachment.html>