gnustep-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnustep-cvs] GNUstep Testfarm Results


From: Fred Kiefer
Subject: Re: [Gnustep-cvs] GNUstep Testfarm Results
Date: Mon, 14 Aug 2006 15:04:54 +0200
User-agent: Thunderbird 1.5.0.4 (X11/20060527)

Hi Greg,

I would prefer to turn the tone a bit down again. What you wrote may be
regarded as very negative by some.

I don't know about any specific GNUstep feature, we did not implement in
the last year because we actively decided that support for gcc 2.95 was
more important. Until such a case turns up, why not go along with what
Adam and Riccardo suggested: We don't offically support gcc 2.95, but we
try to keep things running in that environment. If we know about issues
and know how to fix them we should do so.
I remember that from time to time we break compilation for older
compilers by declaring new variable in the middle of the code. Now I
don't see any problem in only declaring variables at the beginning of a
block. Actually we all try to do this, although it old fashioned by now.

Talking about ancient architecture. Isn't GNustep all about that? We
started off as a step in replacement for NextStep, we just should not
talk too negative about old things :-)

Cheers
Fred


Gregory John Casamento schrieb:
> Riccardo,
> 
>> In any case, and I know I think differently than several people here, I
>> would prefer to retain gcc 2.95 compatibility at least for the core
>> libraries and, if possible, for all "GNUstep supplied" applications. Or,
>> at least, the fundamental ones: gorm, project center, system preferences
>> and if possible gworkspace.
> 
> So basically all of GNUstep, is what you're saying.   This seems to run
> counter to what we're discussing.   The problem is that there are too
> many limitations we must impose on the code in order to maintain that
> compatibility, as I described in my previous posting.
> 
>> Of course this does not apply top linux/x86 which almost everyone uses.
>> It is just for the 1% of the remaining 1%. But it is one of the reasons
>> why I always liked gnustep.
> 
> GNUstep needs to move forward, we can't be held back by the 0.01% of people
> who might be using it on an ancient architecture.
> 
> While we should strive to have a wide variety of machines, we shouldn't
> go out of our way to make it work on machines which are no longer in common
> use.
>  
> --
> Gregory John Casamento
> 
> 
> ----- Original Message ----
> From: address@hidden
> To: address@hidden
> Cc: Fred Kiefer <address@hidden>; Developer GNUstep <address@hidden>
> Sent: Sunday, August 13, 2006 4:26:26 AM
> Subject: Re: [Gnustep-cvs] GNUstep Testfarm Results
> 
> Hello,
> 
> On Friday, August 11, 2006, at 12:02 AM, Adam Fedor wrote:
> 
>> I narrowed it down to one method, but that doesn't really help much. On
>> the solaris, I'm still using the 2.95 compiler, mostly to check for
>> backward compatibility. Perhaps I should just upgrade and start
>> deprecating support for gcc 2.95?
> 
> I used to regularly test on gcc 2.95 the core libraries on various
> platforms. It is just that lately I was too busy with other important
> life tasks.
> 
> In any case, and I know I think differently than several people here, I
> would prefer to retain gcc 2.95 compatibility at least for the core
> libraries and, if possible, for all "GNUstep supplied" applications. Or,
> at least, the fundamental ones: gorm, project center, system preferences
> and if possible gworkspace.
> 
> I do not want to enter the eternal gcc 2.9 versus gcc3.x or gcc 4.x
> discussion. Mountains of ascii bytes have been spilled about that. I
> know 2.95 has many problems. Even Linus Torvalds spoke about the issue.
> I just want to remember that newer gcc's are not a good option on more
> than one platform (mostly because of C++, not because of obj-c or C
> itself) and that gcc guys aren't that happy anymore to fix problems on
> arcane platforms as they did on 2.9 series. Furthermore not everybody
> may want to upgrade gcc (possibly because the box is managed by others
> and 2.95 is the easiest choice) and having more than one compiler
> installed (of which some are incomplete, like lacking C++ or in any case
> have big api differences like 2.95 and 3.x have) is cumbersome.
> 
> Of course this does not apply top linux/x86 which almost everyone uses.
> It is just for the 1% of the remaining 1%. But it is one of the reasons
> why I always liked gnustep.
> 
> Regarding solaris and the bug in discussion, I have gcc 3.x installed so
> we can check this out too. On Sparc but OpenBSD I have 2.95.
> 
> I will resume testing on more arcane platforms ASAP.
> 
> Have fun,
>    Riccardo
> 
> 
> 
> _______________________________________________
> Gnustep-dev mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/gnustep-dev
> 





reply via email to

[Prev in Thread] Current Thread [Next in Thread]