Defining the KDevelop Platform

[Jens Dagerbo]

On 4/6/06, Matt Rogers <mattr at kde.org> wrote:
> Hi,
>
> IMHO, We need to answer the following questions about the KDevelop platform in
> concrete terms. Right now, we don't really have a clear cut idea of what
> the platform actually is. In order to move forward with the idea of
> the platform, i think we need to give it a bit of formal definition.
>
> 1. What will the platform encompass? (don't just say the current
> kdevelop lib dir)
>
> 2. What should the platform be used for?
>
> 3. What can developers who want to use the platform do with it?
>
> 4. Why should developers use the platform?
>
> 5. What advantages would the platform give them?
>

Thanks for bringing this up, Matt.

Questions 2&3 and 4&5 feels like the same question. But I'd like to
offer one more:

6. What's in it for us?
That is, is the "KDevelop Platform" idea of any benefit for the
KDevelop project, or is it just something that will cost us time and
resources to maintain?

I suspect the answer to that depends on the answer to the other
questions. If they can be answered in a way that convinces us (and the
hypothetical users of the "platform") that this is a worthwhile idea,
then we could stand to benefit in terms of additional testing and
potential manpower. I must admit I'm not entirely optimistic.

2&3: What can you do with it?
Unless we aim to be as generic as the KDE framework itself, I think
the answer to that is "an IDE-like application". You're not going to
use the KDevelop Platform (too long. KDP?) to build the next KWord.
You might use it to build KScope. If you do that, you could get a
FileSelector plugin for free.

4&5: So why should you?
In KDev3 you might want to use KDP because you'd get a working (if a
bit confusing) plugin framework. It might also appear that we had
tamed KMDI somewhat.
Come KDE4 and KDev4, KMDI is gone. If everything goes as hoped, Qt
will replace it, perhaps leaving us with not a whole lot new to offer
in terms of UI (unless of course we come up with something new and
fantastic - Alexander's zooming UI? :) ) which leaves the plugin
framework.

Other things that could be in KDP:
#Editor integration. Well.. ok, sure. But it's not that hard to
integrate an editor via KTE as it is.
#Designer integration. Maybe. Can that be done separately from the
language support (which I'm sure all agree should not be in the
platform)? Would anyone want it?
#Generic mimetype based kpart integration. Again, not that
fantastically difficult to do on your own.
#Anything else..? Maybe it's just my limited imagination, but I'm
drawing blanks.


Hopefully the Quanta guys can give us some pointers. What would you
want to see in KDP? What would make it worthwhile for you?


// jens