Re: [c++0x] result_of

["Jonathan Wakely" <jwakely dot gcc at gmail dot com>]

2008/9/30 Paolo Carlini:
> On Sep 19, 2008, at 12:36 AM, Jonathan Wakely wrote:
>
>> 2) According to TR1 all function objects defined by the standard
>> library must give the right answer with tr1::result_of. Making that
>> work might require a result_type typedef or a result_of partial
>> specialisation to be added to some components outside the tr1 dir e.g.
>> std::default_delete (or tr1::result_of could use decltype too.)
>
> To be clear, the usual policy: we want the TR1 facilities to be essentially
> frozen. Certainly we don't want to update / extend the TR1 facilities to
> deal "correctly" with newer components, objects, etc, specified in the C++0x
> draft. TR1 must be consistent with C++03 and only with C++03.

OK - there still might be C++03 components for which tr1::result_of
doesn't work, but I can't think of any right now :-)  C++0x's
default_delete was the first example I found ... there might be no
problems with the C++03 parts.

>> This probably isn't ready to go in, but here's a patch for discussion.
>> If I've missed an obvious way to do it better please point it out! :-)
>
> I think the patch is Ok as-is to close the PR. But please, try to follow the
> iter of the issues which you posted on the reflector and the following
> resolutions: eventually we don't want #if 0 blocks in the code delivered as
> part of gcc4.4.0, much better making a well thought choice and in a comment
> providing pointers to people willing to further investigate the options.

Agreed.  The "reference type" issue will be LWG 904, I'll make sure
the pointer-to-member issue is addressed too, either as part of 904 or
separately.

Thanks,

Jon