On Tue, Sep 26, 2000 at 04:23:04PM +0200, Thomas wrote:
> - I wish that all development with the koffice-libs was in one place because
> that would make it much easier to incorporate changes to all the dependent
> applications. This would effectively mean that the projects theKompany is 
> working on would have to be in kde-CVS. (This is just my personal opinion)

This is a very good point IMHO. The koffice libraries are far away from
being a real "almost-final" framework. In the current state we don't guarantee
binary/source compatibility (with koffice 1.0) . It's not that the current
framework sucks, it's just that it's still too young for third-party
development (IMHO)

That means there will be source incompatible changes in the future
(until a certain point in the framework is reached of course, but I think
everyone agrees that it's not the right time, yet) . When such changes are
made, then it is obvious that the person doing these changes also has to
fix all components, whereas "all" means all components the developer can
access, which effectively means all the components in the koffice module
in the KDE CVS Repository. (like we had last weekend when I removed all
the shell hacks in the components, which required source incompatible
changes -> I fixed all components I had access to -> cvs.kde.org:koffice/* ) .

This is not a nice situation (and in fact it must be a real nightmare for
third-party developers, not having their open components in the same CVS) ,
but I currently see no other way. The framework is just not mature enough.

As soon as a certain point in the framework development is reached and as
soon as we all agree/decide that we're ready to guarantee a source and binary
compatibility (which we will _have_ to do at some point in the future) ,
then of course third-party development is a realistic thing, I believe.

Until then the easiest way to develop a koffice component is to develop it
in the KDE CVS Repository. Then it's for sure that people will fix those
components when doing (unfortunate) source incompatible changes, and in
addition it will (obviously) increase the amount of beta testers for the
specific component. Everything else is probably a nightmare for the
developer?

Just my humble opinion.

Bye,
 Simon