Re: {IMPORTANT] README, Sawfish 1.6.0

[Janek Kozicki <janek_listy wp pl>]

Christopher Roy Bratusek said:     (by the date of Tue, 10 Nov 2009 14:40:00 +0100)

> - Tabs ... is anyone willing to work on them?°

I can't :/ It would be quite nice if somebody wanted to fix the tab
stacking bug. With this bug you can see a window "halfway through"
tabs in another window.

> - Merge stuff from sawfish-mmc ... @Michal: you said to provide some
>   patches ... any news?°

oh yeah... -mmc is waiting quite long :)

> Patches to review, I want to see them inside 1.6.0, but I can't test
> Multihead on my system, so I depend on you at this point:
> 
> Xinerama aware grow-pack [1]*
> Xinerama and XConfigure Window [2]*
> 
> [1] http://sawfish.wikia.com/wiki/Xinerama_aware_grow-pack.jl

I tested it. It could be improved, but this can be done even after
release. It is so much better than nothing, you wouldn't believe it :)


> [2] http://sawfish.wikia.com/wiki/Xinerama_and_X_ConfigureWindow_fix

This definitely has to go into release, because:

- without this patch xinerama is simply unusable, I'm losing a window
  once per minute.

But this patch definitely needs more work, testing and improvements.
I'm afraid though that 6 weeks isn't enough to reach a 100% fix of
those problems. With current patch I reduced the bug by roughly 95% of
original frequency. And that's why it should go into release, but
also, it still needs more work.

I'm pondering whether simply *deleting* completely this error_handler
would fix the problem? I think that maybe we don't need it at all,
because when a window gets an error and X closes it, it is sending also
a standard event about "This window is closed by X now, Event". And a
regular remove_window gets called.

I'm pondering that maybe error handler tries to be "faster in closing
windows" than really necessary.


> 22nd Nov or 4st Dec? (+/- 12 days)

Let it be 22nd, as you wish :) I cannot adjust to the timeline
anyway, because my RL work is so occupying.

 
One more point, I consider it critical: Mouse buttons. If we make a
release 1.6.0 with evdev, as it is currently - we will be bashed into
oblivion by complaining users that we have never heard from before.
I tell you, bear my words. People will subscribe to ML only to complain.

* All current bindings to Button6 7 8 9 (and therefore more bindings
  from evedev: Button10 11 12 13 ..... 20 or infinity(?)) have to
  work as they worked before. They must not be ignored. Who wants to
  redefine 100+ of his custom bindings? (by editing some
  obscure .xbindkeysrc and other stuff ?)

If we cannot reach this goal currently then Mouse Buttons patch
should be reverted. Better have 9 nicely working buttons than
crappy/broken support for 20 buttons.

* Or we could make "evdev"/"9 Buttons" support optional, either
  though compile time flag, or better through SawfishConfig.
  People will enable it, when they want to.

Hold on.. what was the problem with russian keyboards? Did 9 buttons
broke russian keyabord mappings?

best regards
-- 
Janek Kozicki                                                         |