Re: [MiNT] XaAES (fwd)

[Standa Opichal <opichals@seznam.cz>]

Again moving the private discussion to the mailing list...

STan

---------- Forwarded message ----------
Date: Wed, 19 Mar 2003 15:37:02 +0100
From: olivier landemarre <olivier.landemarre@utbm.fr>
To: Petr Stehlik <pstehlik@sophics.cz>
Cc: Frank Naumann <fnaumann@cs.uni-magdeburg.de>,
     Standa Opichal <opichals@seznam.cz>
Subject: Re: XaAES
Resent-Date: Wed, 19 Mar 2003 15:35:18 +0100 (CET)
Resent-From: OpichalS@seznam.cz
Resent-To: stop@idoox.com

A 14:52 19/03/2003 +0100, Petr Stehlik a écrit :
>On St, 2003-03-19 at 14:44, olivier landemarre wrote:
> > here XAAES looks a little bit strange, I think. I think too the driver for
> > mouse is slow and need a lot of ressource for nothing.
>
>I don't remember having problems with moose.xdd.

No problem with this driver, but slow. The problem come in a part XAAES
use the xdd and for an other part use vq_mouse() this is very strange because
some old event can be process after have before information with vdi.
I notice only that this driver is slow and could be more direct, ok it's 
look more Unix like
but is it need, no xdd look work fine.


> > Other problems are not solve and new dev doesn't work at all, there will
> > have big rewrite in mouse use and it's a little bit hard
>
>"new dev"? Are you writing new mouse driver or what?
No absolutly not, because it should work perfectly with the actual one, and 
it's not need. But
I wan't erase access to mouse with vdi function. This is already done but 
not work (it seem
in fact coming with this change because I restore vq_mouse() and I have 
still alway's same problems)
So I suppose the problem come with my patch for bug of MU_TIMER use in 
evnt_multi() (it's use as evnt_timer()!)
it look easy to patch it but I have done somewhere a mistake and I can't 
test in same time because my winAranym can't access to local filesystem 
under win98 and I have no way more for test except at work the day after!

>Big rewrite in mouse use? Why? I am very happy with xaaes being able to
>serve AES events even if mouse stands in a pull down menu. Why are you
>going to rewrite mouse use?

Because there is bugs, not in driver, but in use, it's easy to have this 
sort of problem for example if you do yourself popup menu. Generaly you 
have clik one time so you display menu, if you do this faster, after this 
you wait about position cursor and button evnt with evnt_multi() and here 
you have the first bug, when you do this first time evnt_multi() ask state 
of mouse with vq_mouse(), and not wait for an evnt button from mouse 
driver, so if you not depress button enough faster application get a bad 
button down and popup close selected I don't know what item! I think it 
easy to solve this problem (I'm sometime to confident!) so I cut the test 
of vq_mouse(), put in XAAES event cue, and I thought problem will be solve 
with only this and it was true! now popup work perfectly, but now some 
applications working before doesn't report click button anymore sometime 
and I don't understand why?!!! So I have decide to revue most of mouse 
gestion.


>Shouldn't this be discussed somewhere first? These major changes without
>prior explanation scare me a bit.
The problem of mouse is the most important I think in XAAES because this 
soft is well write I think and easy to understand


>Petr
Olivier