Re: [PyWM-talk] [Fwd: Re: My initial pywm fumbling and a window handling bug]

[Nils O. F. <ni...@no...>]

Elmo, thanks for the reply

Torsdag 04 januar 2007 16:03, skrev Elmo M=E4ntynen:
> Nils Olav Fossum wrote:
> > hi to all 11 of you :D

> > ..but there is a bug in the window-handling code somewhere..
> > The windows dont register as they should do.
> >
> > in the __init__.py file, WM class, we have _addWin, _delWin and window
> > I had to manipulate these to get the window registering working.
> > See the attached patch.
> > This is just a hack so I can go on testing pywm, but I hope someone can
> > take a look at it to locate the real bug..
> >

Just some more meat in case someone out there stumbles into the same:
My change to WM._addWin are just a harmless check so new windows dont=20
get 'registered' twice.
The WM._delWin change fixes a bug in releasing the window from=20
the 'WM.windows' dir.
The real workaround is done in WM.window.

The underlying problem is that WM.window is called=20
(from the code in flwm_.pyx ?) before WM._addWin 'registers' the new window.
Here is a printout of wm1.py after the patch is innstalled:

=2D------
wm.on_keyEvent: known key, launching func
Oooopss, should not happen?? (window added anyway)
  **** Cheated to add window, it was not registrered with _addWin! ****
WM.on_activate: activated 'rxvt'
Whooha: window already exist !!! (skipped add)
pWin: <window hWin:0x81cf890 'rxvt'>
WM.on_create: created window 'rxvt'
=2D-----

Without the patch I get a exception error.

BTW, wm1.py was the smallest app I could come up with to test this bug (and=
=20
the one below).=20


> >
> > So a question..
> > I use Xnest for testing pywm because a 7" 800x480 sized screen is less
> > than comfortable to work on.
> >
> > When I start my pywm derivate (wm1.py), on a new Xnest with no
> > applications, something goes wrong and I cant get my test Fl_Window to
> > receive mouse click. Started like this:
> > $ Xnest -ac -geometry 800x480 :22 &
> > $ export DISPLAY=3D:22
> > $ ./wm1.py
> >
> > However when one application is running inside Xnest before I start
> > wm1.py it works.. (Started like this):
> > $ Xnest -ac -geometry 800x480 :22 &
> > $ export DISPLAY=3D:22
> > $ xclock&
> > $ ./wm1.py
> >
> > So , is this another bug, or is it something  I dont do right..?
>
> Probably a bug. Can't tell, since I got too much on my hands at the momen=
t.
>
Oki, I can live with a workaround while testing a bit more.=20

> You should consider using trying the darcs vcs. The tarball is a darcs
> repo (a branch), with which you can work with (it works without a
> central server, but I could set one up if deemed useful). Notice that
> your patch has tabs in the last part. Also, you might want to take a
> look at Leo, the editor (or ide) that the original author used. Because
> of the learning curve (not that steep) associated with efficient usage,
> I haven't used it much, but if you if like the style of workflow it
> enables, it might nice.

Thanks for the tips, Im looking into it.
Sorry about the tabs in patch..=20
I will send a proper one if asked.

> So, I haven't worked on pywm much, and can't at the moment, but I do
> have plans concerning it. Among other things, I've considered rewriting
> or adapting it to use the (hopefully) to be python interface to the new
> x lib.=20

Great, I really hope theres a future for pywm!

> But for now, do with it as you please. I'd be glad to hear from=20
> big improvements though, and if you continue to develop it, it would
> nice of you to tell about them on the webpage.
>
> Elmo, the third (named) maintainer

heh, dont expect too much from me. Im not a seasoned programmer :)
pywm just seem to be a good tool for me to controll some applications in a=
=20
touchscreen environment.
I can navigate around in python code, but when it comes to C Im lost.
Therefore I really hope theres no showstopper bugs in the underlying flwm=20
code.

I will keep you posted while I continue to fumble with pywm.

:)

=20