[cfe-commits] r64485 - /cfe/trunk/docs/LanguageExtensions.html

[Chris Lattner]

On Fri, 13 Feb 2009, Eli Friedman wrote:
> On Fri, Feb 13, 2009 at 12:00 PM, Chris Lattner <sabre at nondot.org> wrote:
>> +<p>The result of __builtin_shufflevector is a vector
>> +with the same element type as vec1/vec2 but that has an element count equal to
>> +the number of indices specified.
>> +</p>
>
> clang currently enforces the constraint that the element count of the
> result is the same as the element count of the inputs.  Should we
> remove that constraint?

LLVM IR on mainline allows the two input vectors to be different types and 
the output to be different types.  We should allow this at some point.

-Chris

-- 
http://nondot.org/sabre/
http://llvm.org/