Re: [Gnustep-cvs] GNUstep Testfarm Results

[Fred Kiefer]

Adam Fedor schrieb:
> On 2006-08-11 04:27:51 -0600 Fred Kiefer <address@hidden> wrote:
>> I agree with all of you that we should not waste to much effort on
>> getting it working, but in this specific case, we have just one method
>> that is failing and we know that up to a certain date this wasn't the
>> case. Perhaps it is not too hard to sort out this method?
>> Adam, which one is this method? And are you willing to test different
>> implementations? BTW which version of gcc 2.95 is installed, 2.95.2 or
>> 2.95.3?
>>
> 
> GSString.m compiles if I take out these methods (in two different classes)
> 
> - (NSString*) substringFromRange: (NSRange)aRange
> {
>   GS_RANGE_CHECK(aRange, _count);
>   return substring_c((GSStr)self, aRange);
> }
> 
> - (NSString*) substringWithRange: (NSRange)aRange
> {
>   GS_RANGE_CHECK(aRange, _count);
>   return substring_c((GSStr)self, aRange);
> }
> 
> ======
> gcc version 2.95.3 20010315 (release)
> 
> 
> I haven't really looked into when it changed.
> 

In the most recent GSString.m version these methods are there only once.
The first one is obsolete anyway, see NSString, where it gets defined as
calling the second one. As this in now the only remaining usage of
substring_c, we could try to do the inlining ourselves and put the code
from the function inside the method. If this alone does not resolve the
compilation issue, we could try to rework the actual code. Here the
interesting question could be why the similar which calls substring_u works.

Fred