Future of KDE Development

[Lauri Watts]

On Friday 18 February 2005 06.43, Aaron J. Seigo wrote:
> On Thursday 17 February 2005 10:08, Stefan Teleman wrote:
> > which works on most platforms. my understanding is that for KDE 4, BC
> > with KDE 3 needs not be maintained. so, i am offering to fix the
> > things which don't happen to work on Solaris. but, there's also Linux
> > and FreeBSD. so, i was asking if, in case we run into a situation
> > where something works on Linux but not Solaris, or FreeBSD, or
> > vice-versa, could we work together to try to find a common solution,
> > rather than relying on separate source code patches. when necessary,
> > by abstracting the OS-dependent interfaces, for example. i do realize
>
> i think this has always been the desire and goal, we've just lacked people
> to act as intermediaries between some of these pools of downstream patches
> and the official KDE sources. if you're willing to do this for Solaris,
> then that would be awesome. all it takes is for someone on Solaris to post
> a patch saying "this makes it work, i'm not sure it breaks on other OSes"
> to get feedback, fixes.

You know, when we post patches for FreeBSD, it's usually a handful of the 
usual suspects that will test them, if anyone does at all. We get much 
quicker feedback by just committing them.  If it breaks Linux, someone is 
pretty quick to tell us (not that we do so intentionally, any more than the 
majority of developers go out of their way to break FreeBSD).  

Additionally, there have been several cases of patches to make things work for 
us actually fixing the same issue Solaris, or Tru64 or whatever, and vice 
versa.  A lot of what we do is not so much 'make it go on FreeBSD' as 'make 
it not Linux specific', and that helps everyone out long term.

So, a small plea: If you see these kind of patches and you are *not* using the 
platform in question, it's still very useful to have them tested before they 
go into CVS.  Knowing what it doesn't break, is just as useful as knowing 
what it does, maybe more so.

Stefan: I'm going to guess that like the changes we make, the majority of them 
fall into a small handful of common fixes (for instance, adding headers that 
are included implicitly on Linux, working around glibc'isms and libs like the 
gnu math libs, busted configure checks and dealing with hardcoded paths that 
oughtn't be, and the occasional unbashifying of a script)  We (well, mostly 
Adriaan) must have fixed pretty much the same timezone miscalculation bug 
about 34 times over, sometimes multiple times in the same app as it is 
rewritten and the previous error creeps back in.    I'd be very surprised if 
a whole lot of your patch set isn't the same handful of things, over and 
over.

Maybe we should start writing the common cases and their fixes down somewhere, 
so we have a resource to point developers to if we find we're fixing the same 
problems over and over.  Education does work, so does being visibly involved, 
and pushing things upstream as soon and as often as possible.

Regards,
-- 
Lauri Watts
KDE Documentation: http://docs.kde.org
KDE on FreeBSD: http://freebsd.kde.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20050218/f9ab8e51/attachment.sig>