On Fri, Aug 01, 2003 at 09:09:53AM +0200, Dominik Vogt wrote:
> On Thu, Jul 31, 2003 at 10:45:12PM +0200, Olivier Chapuis wrote:
> > On Thu, Jul 31, 2003 at 04:44:48PM +0200, Dominik Vogt wrote:
> > > On Thu, Jul 31, 2003 at 03:47:56PM +0200, Olivier Chapuis wrote:
> > > > On Thu, Jul 31, 2003 at 11:51:59AM +0200, Dominik Vogt wrote:
> > > > > Somethimes I feel like talking to a wall. See my previous mails.
> > > >
> > > > I think that you do not understand me. I see bugs that I can reproduce
> > > > easily and I want to found a solution. I really care that fvwm can
> > > > work well with certain applications (e.g., Mozilla and gkrellm). And
> > > > what about all your work on autohide with the Schedule command and
> > > > FvwmEvent (via WindowShade)?
> > >
> > > I'm talking only about not grabbing the pointer in complex
> > > functions. Argue as much as you want, it does not work because it
> > > can not work reliably under X, for example because of the way
> > > EnterNotify and LeaveNotify events work.
> >
> > Please give an example I can reproduce. I think there is some but I would
> > like to see one which is reproducible. The example you give cannot
> > happen I think (and there are problems in the code with this example).
> > Note, again, that aborting complex function can _also_ break things!
>
> There have been several examples in the past. Search the mailing
> list for them and use the old tarball they appeared in. Please
> don't waste my time generating artificial buggy situations for
> you.
>
The only example I found is the well known "rxvt selection problem"
and ... it happens on "H"old. We are not agree on a problem it seems
to me that you can make an effort and give me an example (as I said I
worked hard on this, this last few days). Or should claim that there
is no to see one?
> > It is clear that you can do too much things with complex function to
> > have a default regarding grabbing. My hope is/was that this can be
> > detected in the average and add a special stuff for the unlikely
> > case.
>
> It can't.
>
> > I think that Dan idea is ok (I do not want to fight too much
> > more).
>
> I don't. Sometimes it's better to live without certain "features"
> for the sake of code maintenance.
>
There is no reason for a function of the form
AddToFunc I FvwmEventHandle_config_window Echo "config_window"
to be aborted because the pointer is grabbed. Again a complex func can
do so many things that controling the grabbing is necessary IMO.
> > I've suggested the converse: added a ForceGrab and
> > ForceFastGrab "dummy command" because again I cannot reproduce
> > (or even get one problem with my logic).
> >
> > > If you don't like it,
> > > all you can do is to delay function execution until the pointer
> > > can be grabbed (i.e. loop forever in certain places). But - this
> > > may generate a deadlock with applications that grab the pointer
> > > and then expect that the window manager processes requests.
> > >
> >
> > Do not underestimate my understanding of fvwm and X (however, I am not
> > (yet?) a big specialist of complex functions).
>
> > In a certain sense fvwm already "freeze" when grabbing fail: when
> > grabbing fail when you execute a complex function fvwm freeze for a few
> > seconds in GrabEm and in fact this cause the problem with gkrellm: If
> > you try to grab only once when executing complex function and abort if
> > this fail gkrellm can be moved smoothly.
>
> It's a gkrellm problem. Period. Not every bug in applications
> can be circumvented by the window manager.
>
Yes it is a gkrellm problem (and maybe a gtk 2.2 problem). The problem
I've with my "old" gkrellm come from FvwmEvent. What about the others
examples I give with FvwmEvent?
> > > > You give an example and I study it carefully and found some strange
> > > > behavior in the current code.
> > > >
> > > > You claim that my ideas can give raise to some bug. I cannot reproduce
> > > > any one but I trust you.
> > > >
> > > > I say that I am ready to work on any other idea.
> > > >
> > > > Now you say that I am a wall! This is funny. Maybe you should read my
> > > > mail
> > >
> > > I spend much more time reading list mail carefully than you think
> > > (and than I should spend, for that matter).
> >
> > My previous and _complex_ previous long mail as the following date (I
> > spent one day to write it, reading the code an your mails, make some
> > experience ...etc):
> >
> > On Thu, Jul 31, 2003 at 11:32:05AM +0200, Olivier Chapuis wrote:
> >
> > Your answer arrived 20 minutes later ... :o)
>
> Well, good job in writing that mail. Although it was long, it was
> easy to understand.
>
My feeling is that you do not want to do something (on this subject)
whatever the problems we can discover with the current code, whatever
I can or any body else can say and whatever the solution we can propose
(an optional pre cmd) ...etc. So yes it is easy to answer my mails.
Regards, Olivier
--
Visit the official FVWM web page at <URL: http://www.fvwm.org/>.
To unsubscribe from the list, send "unsubscribe fvwm" in the body of a
message to majordomo_at_fvwm.org.
To report problems, send mail to fvwm-owner_at_fvwm.org.
Received on Fri Aug 01 2003 - 04:55:58 BST