[llvm-commits] [llvm] r125595 - in /llvm/trunk: lib/MC/MCParser/AsmParser.cpp test/MC/AsmParser/exprs.s test/MC/AsmParser/paren.s

[Bill Wendling]

On Feb 23, 2011, at 12:32 PM, Joerg Sonnenberger wrote:
> On Wed, Feb 23, 2011 at 12:06:52PM -0800, Jim Grosbach wrote:
>> On Feb 23, 2011, at 11:54 AM, Joerg Sonnenberger wrote:
>> 
>>> On Wed, Feb 16, 2011 at 09:47:22AM -0800, Jim Grosbach wrote:
>>>> What about other targets? ARM, MIPS, etc.. shouldn't accept
>>>> bracket-as-parens expressions unless that's part of their documented
>>>> syntax.
>>> 
>>> Can you name a specific platform for which GAS doesn't accept it?
>>> 
>>>> For example, the following is not legal ARM assembly and should give a
>>>> syntax error on the '[' token.
>>>> add r0, r1, #[_foo - _bar]
>>>> 
>>>> With these changes, that's not happening anymore.
>>> 
>>> _foo:
>>> _bar:
>>> 	add r0, r1, #[_foo - _bar]
>>> 
>>> is accepted by GAS and gives the same result as the expression with ().
>> 
>> 
>> You are operating under the mistaken impression that what gas accepts
>> is the definition of what is legal assembly. This is a bug in gas for
>> it to accept that expression for ARM. It may or may not be a bug for
>> it to accept it for X86. I don't know X86 assembly syntax well enough
>> to know the answer to that.
> 
> Since you seem to operate under the impression that you are right and I
> am wrong, feel free to change it however you like as long as it doesn't
> break the x86 backend.
> 
Jim demonstrated how this breaks ARM. Therefore, it's not just his "impression" that you are wrong, but you *are* wrong.

-bw


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20110223/fc58bc08/attachment.html>