Why modules? (was: Re: A humble packager's request: Retire kdeaddons.)

[Chris Cheney]

On Thu, Feb 12, 2004 at 12:05:19PM +0100, Rob Kaper wrote:
> On Wed, Feb 11, 2004 at 10:19:57AM -0600, Russell Miller wrote:
> > Why do we have modules at all?
> 
> So that developers and early adopters won't have to check out several
> gigabytes of kdepim+kdenetwork+kdegraphics when they just want
> libs+base+games+multimedia. And we can't just put every application in its
> own module either, because that would pretty much remove the E from KDE.[1]
> 
> We *do* inconvenience ourselves (and sometimes packagers, I suppose) by
> pre-organizing our software. But that's the entire point. We're not one big
> heap of code and we're not countless seperate entities either: we're
> organized. That means we have fights on mailinglists and occasionally hate
> each other so much that we need weeks of therapy in remote Czech locations
> to work together.

organized? *cough* hah *cough*

KDE inconveniences packagers greatly by pre-organizing software. There
is no way possible that I can stress greatly enough. In an ideal world
all of KDE would be busted up into the individual apps and library
sources in KDE CVS.

> But in the end, we can tell the world, "here's KDE x.y".
>
> > All this separate module stuff seems to be causing a lot more trouble than
> > it's solving.
> >
> > This would remove dependency problems, remove squabbles over where to put 
> > particular software packages, and solve the problem of duplicate compiles 
> > (each program would be able to build the dependencies that it needs).
> 
> But it would demote us to a bunch of hackers saying "here's hundreds of
> applications, have fun turning that into a desktop environment". KDE is more
> than that.

That would actually be *much* simpler than the way it is now, at least
for Debian. It would also mean that Debian might actually be able to
ship with a modern KDE once in a while.

> [1] We *could* actually *develop* less centralized, but prior to release it
> is vital we pull the bits together for peer review and integration testing.

Developing in separate source trees and having peer review is not
mutually exclusive and would probably lead to less buggy code in the
long run.


Chris
Debian KDE Maintainer
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20040217/289731b3/attachment.sig>