[Tuxpaint-devel] hello, and hildon

[<ter...@ko...>]

On 2/24/07, Kees Jongenburger <kee...@gm...> wrote:
>
> Hello Terje
>
> It looks like we are enough people who have a good interest in this
> project
> the mud builder builds tuxpaint from the tar.gz archieve.


What are the advantages of using mud-builder instead of own packaging? As
far as I've understood, mud-builder is great at building simple stuff with
no Maemo specific changes. When the building has to be done in a customized
way to make tuxpaint even run on Maemo, mud-builder is maybe not the optimal
solution.

I just tried to build tuxpaint from the debian repo and it required
> hevea and such for building
> how did you build it? I am quite sure the source debian will need to
> changes , the question is where to put these changes :p. I am off to
> fosdem are there tuxpaint developer going there?


I suggest we use the garage project Alessandro started to make the Maemo
specific changes. Any changes that need to touch the tuxpaint code, should
be committed to tuxpaint's CVS. Stuff like Debian control files, which are
Maemo specific, could be hosted in the garage project's svn.

There are at least two different types of tasks to do:

= Packaging =
It's probably easier to start the packaging based on a stable version, so we
could start with tuxpaint 0.9.16's source tar.gz package, make the needed
changes to get it running well on 770 or N800, and then see how the changes
could be merged to tuxpaint CVS. Probably the changes done on 0.9.16 are
pretty easy to port to tuxpaint CVS.

Maemo requires some special dependencies for packaging, so as an answer to
Caroline, we cannot push that to Debian's tuxpaint package. In addition,
Maemo is using older debhelper and some other tools.

= Hildonizing =
I'm mostly worried about the mouse cursor. Otherwise I guess hildonizing
will not be a big task. Mostly replacing some gtk calls with hildon calls.
I'm not sure how file system differences etc. are going to affect.

Hopefully most of the stuff can be done without touching the generic
tuxpaint code. Compile time flags are not an optimal solution, but I guess
they will do for now.

How does that sound? I'm sending Alessandro a please-let-me-in email to get
this started. I guess it's still best to keep the discussion in this mailing
list.