Re: [pygame] New GUI

["Patrick Mullen" <saluk64007@xxxxxxxxx>]

> Show me a GUI builder that in fact produces clean, non-overloaded code,
> that does not break, if I manually change, add or remove stuff and so
> effectively eases development. I am not aware of any. If they do not
> match these criteria, but I am instead forced to stick with them (like
> e.g. VS.NET), they are useless.

Gui builders really ought to store the gui structure and then let the
coder write all the functionality hooks.  Like, oh, I don't know,
writing html.

Oh wait, even html designers get it wrong.  Oh well :)

Yeah, gui builders aren't all they are cracked up to be.  I am using
CEGUI for my project, and I used the gui builder for a while, until I
discovered that it randomly chose whether positions were stored as
percent of screen size or pixel size, and I had to redo the whole
thing manually.

Don't criticize Marcus for not talking to other gui developers about
their gui's problems, as far as I know, he was the first :)

For me, I probably wouldn't purchase a gui toolkit.  I have written
some in the past, and find writing guis rather enjoyable.  If I want
to actually work on a game rather than a gui, I will look at what's
out there and use it if I feel like it.  Both PGU and ocemp really
look like they control how I would write the game, which I think is
fine in PGU's case, it's meant to be sort of a quickstart, not a last
minute plugin.  CEGUI definitely has influenced how I have programmed
my game, but it has also saved me some time.

If I couldn't find a suitable free library though, I would probably
write my own toolkit as you have just done.  Trolltech has succeeded
with QT because is is usable in so many places for many different
operations, and is also used in one of the two most popular linux
desktops: KDE.  If it weren't for their liscensing though, I can't
help but think that GTK would not have done as well as it has.

For a toolkit that can only be used in pygame, ignoring things like
pyopengl, python-ogre, etc, such limited use I don't see as being very
marketable.  If it was generalized, and pluginable to almost any
python program, that starts to have some real value.