Re: NEWS on 1.6.0, crash fix... semi-fix (if you create and quickly delete a window sawfish will segfault)

Am Donnerstag, den 08.10.2009, 00:52 +0300 schrieb Timo Korvola:
> On Wednesday 07 October 2009 16:27:48 Christopher Roy Bratusek wrote:
> > I tried about 25 times with your patch applied, and sawfish did not
> > crash.
> My approach there is not mutually exclusive with Janek's, which 
> consisted of adding more error checking to add_window.  To test my patch 
> one should try to trip Sawfish on errors as hard as possible, so it is 
> best not to add any error checking before gaining reasonable confidence 
> that my patch works.  However, eventually some more intermediate error 
> checking in add_window would perhaps be good, although not quite as much 
> as in Janek's patch should be necessary.

Well as I said rox-filer each 3rd or 4th time made sawfish crash on my
system, now after 25 times it didn't happen, so sawfish is now at least
much more robust and I'm somewhat sure Janecks test tommorow will also
be positive. Of course excessive testing would give the final result but
sawfish not crashing with real-life behaviour is enough atm.

Major improvements on X code are planned for 3.0 (I'll extra study X11
for sawfish 3.0).

> > So I merged with timo/race branch (and btw with some others of
> > your branches, too).
> Hm, ok, I suppose.  Some of those other branches were work in progress, 
> or more realistically work without much recent progress. 

> For instance 
> the mouse branch just removes badly implemented support for more than 
> five mouse buttons, but I have not had time to think about a better 
> implementation.  

Well users can use evdev to use button 6 - 13, so they won't loose
functionallity, addtionally I wanted to clean-up a bit, (ugly
mousebutton implementation dropped, gnome-specific code trimmed down,
ancient built-in audio support code dropped, libgnome* stuff dropped
(the later already in 1.3.5/1.5.0 series)) as there's already enough
suboptimal code in sawfish. keys.[ch] for 3.0 should be (partially)
rewritten to match the following:

a) automap AltGr and such to the correct key (instead of Mod?)
a.1) Compose should be named Compose not Multikey (...)
b) differ between Num_Lock(On)-Keypad* and Num_Lock(Off)-Keypad*
c) work correctly with two kbds with different layouts
d) on my system "Alt" is bound to Meta and WM_Modifier (the former via
X.Org the later via Sawfish), so W- and M- bindings work, but A- *don't*
that's inacceptable, Alt is Alt so that has to work. Well Driver is kbd
with pc105/de modell, I'll try later with evdev but either way that
should work in both cases correctly
e) standalone Escape should not be bindable, as it's used to exit from
Compose or nested-keybinding sequences (SawfishConfig allows that)
f) why can't I bind ° key ?
g) more?

Enough on keys: I did some testing with your other branches:
- kde4: merged manually earlier - reasonable changes
- fdo-fix: better code and works perfect here
- frame-fix: better code and works also perfect
- mouse: see above
- race: see above

It's O.K. I think. Either way time for new features is until 6th Dec for
fixes until 20th Dec. Though the former date might be eariler, if Teika
needs more time for Docs (eg. 1st Dec).

> But those branches should be fairly stable.
> One thing went wrong though: you should have merged focus-revert instead 
> of kde-menu-focus.  Now you actually cancelled the changes in 
> transient.jl of commit a2117caef, where focus-revert was merged 
> manually.



Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]