On 21 February 2014 09:15, David Woodhouse <[email protected]> wrote:
> On Fri, 2014-02-21 at 09:11 -0500, Rafael Espíndola wrote:
>> The other issue that has been reported is people doing "clang -S
>> foo.c" with invalid assembly, and this lets them use exactly the
>> option to avoid it.
>
> You're not suggesting that we should use -no-integrated-as *with* -S,
> are you?

When you don't want the integrated assembler, yes.

> That doesn't make sense, because '-S' means "no assembler".
> So -no-integrated-as should be *implicit* in that.

I disagree. Clang can parse assembly and we should make more use of it
than we do now. Right now the only thing we do is diagnose errors
earlier, but we should really be doing things like checking provided
constraints against the instructions.

Cheers,
Rafael

_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to