[cfe-dev] Proposal: parsing Qt signals/slots with attributes

[Douglas Gregor]

On Sep 28, 2011, at 11:34 AM, Sean Hunt wrote:

> On Wed, Sep 28, 2011 at 13:37, Douglas Gregor <dgregor at apple.com> wrote:
>> I had completely forgotten about these. How much work is involved in making attribute namespaces work?
>> 
>>        - Doug
> 
> It depends on whether we want them to work or work nicely, and whether
> or not we want them for GCC attributes. If we just want them to work
> and we don't care about GCC syntax,

Aye, there's the rub. We want to stick with GCC attribute syntax for these attributes, since we don't want to require C++0x. I guess we could talk with the GCC people to see if they're planing to pull attribute namespaces into GNU attributes, in which case we could do the same. That would be fine.

> stuffing special-case parsing into
> ParseCXX0XAttributes would work fine. If we want them to work nicely,
> we would have to put some more work into the TableGen to use the
> Namespaces attribute on Attr to generate some code.


As much as I'd like to get the general solution, time restrictions may force us into special-casing this.

	- Doug