[MAJOR BUG] g++ 2.95.x generates invalid code for KDE 3.3.0

[Thiago Macieira]

Leo Savernik wrote:
>> 20:42 < pme> The bug is already fixed in 3.x.  It's not worth the
>> volunteer time it would require to fix it again for 2.x.
>
>Again? So it was fixed in 2.x and then regressed??

Again as in "more than once". It has already been fixed once, for 3.x.

>> ...
>> can't say it's official gcc devel statement, but it surely means
>> something.
>
>Well, it only states that they have no interest/manpower in fixing old
>releases. But I've never implied that gcc-2.95 should be fixed, but that
> KDE should refrain from using features that break gcc-2.95. That would be
> the easier route.

I don't suppose you want to keep compatibility with g++ 2.7.2.3 as well? We 
sure could do that... We could just revert to KDE 1.1.2 and start from 
there, which, by the way, compiles a lot faster than current KDE 3.3.x 
code.

And I repeat what I said: I see no use in knowingly continuing to use bugged 
software. Either fix it, or upgrade. That's how it goes for every project 
and program out there. Why should it be any different with the compiler? 
*Especially* the compiler should be something you can rely on.

So, our situation is: 
- gcc devels don't support gcc 2.x releases anymore
- we cannot fix the bug in their code (even if we did, I somehow don't think 
it can be done in a binary-compatible way)
- why should we, as in KDE developers, continue to support something like 
that?

What I am proposing is this:
1) for the remainder of KDE 3.x releases, our policy be
	- we recommend g++ 3.3.x or 3.4.x
	- we recommend against g++ 2.95.x, but we make our code work on it as best 
as we can

2) for the KDE 4 release, our policy change to:
	- we (hopefully) recommend g++ 4
	- we support g++ 3.3.x and 3.4.x in the form of #ifdef'ed directives
	- we do not support g++ 2.95.x -- people using it fix the problems on their 
own

Of course, that is not saying that we won't apply reasonable fixes if 
brought to our attention. To paraphrase your reason for using g++ 2.95, I'd 
much rather devote my time to develop than to work around compiler bugs.

-- 
  Thiago Macieira  -  Registered Linux user #65028
   thiago (AT) macieira (DOT) info
    ICQ UIN: 1967141   PGP/GPG: 0x6EF45358; fingerprint:
    E067 918B B660 DBD1 105C  966C 33F5 F005 6EF4 5358
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20040922/0154d7dd/attachment.sig>