[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc-patches
Subject:    Re: [objc-improvements-branch]: FYI: Yet another round of ObjC++ work committed -- ADDENDUM
From:       Ziemowit Laski <zlaski () ziemas ! net>
Date:       2004-06-30 22:07:42
Message-ID: E8C60E8C-CAE1-11D8-A9E3-00039390FFE2 () ziemas ! net
[Download RAW message or body]


On 30 Jun 2004, at 3.10, David Ayers wrote:

> Ziemowit Laski wrote:
>
>> Oh, yeah, and I cleaned up <objc/...> headers to be usable with 
>> ObjC++.
>>
>
> Did you simply merge in Andrew's changes to libobjc or were there other
> changes necessary?

The latter.

>
>>>
>>> David, if you want to whip-up an identifier cleanup patch for
>>> objc-improvements-branch, go for it. :-)
>>>
>
> Here is the cumulative patch adapted for the objc-improvements-branch 
> for:
>
> [RFA] c-parse.in unify ObjC tokens
> http://gcc.gnu.org/ml/gcc-patches/2004-06/msg00782.html
>
> [RFA][objc] Identifier cleanup
> http://gcc.gnu.org/ml/gcc-patches/2004-06/msg01561.html
>
> [RFC/RFA] [ObjC] Class <protocol> support I
> http://gcc.gnu.org/ml/gcc-patches/2004-06/msg02496.html

I'm currently doing a mainline->objc-improvements-branch merge, which 
probably will pick up some of this.  When I'm done, I'll check your 3 
patches to be sure.

>
> I see that you have dealt with gcc/testsuite/objc.dg/id-1.m in a
> different way already on the branch.  But I must admit that I find the
> new diagnostics...
> id-1.m:4: error: conflicting types for 'id'
> id-1.m:0: error: previous declaration of 'id' was here
> (note the missing location and line number 0 as this seems to be a 
> built
> in type now) a bit misleading.

Well, yes, it's a built-in type, but at the same time we must allow 
users
to (re)define it as well.  Certainly, if there is some trivial special 
sauce that
we could apply so that the 'id-1.m:0: ...' message becomes more 
informative,
I have no objections.   But I don't see a point in spending too much 
time on this.  :-)

>
> Anyway, bootstrapped and testing passed (including obj-c++)
>
> Cheers,
> David
>
> PS: Why did you choose 'obj-c++' over of 'objc++' in various places?  I
> seems needlessly inconsistent.

This is primarily to keep gengtype and DejaGNU happy; these dislike it
if the name of one language is a proper prefix of another. :-)

--Zem

[prev in list] [next in list] [prev in thread] [next in thread]