Maybe we should revert 'Iconify' patch.

Hi, dear sawfish developers.

I'm sorry for Ian Zimmerman, but I think the 'iconify' patch
commitment should be undone, in rev4373.

But remember, don't jump on to the quick conclusion. Read and
confirm my opinion.

First, there had not been a bug. Putting 'Iconify -> no' in the
configurator GUI is likely to do *nothing*. This is the same pitfall
I once was in:
(Pretext: I didn't forget this patch, but I don't have time to restart 
it. ;)

# Ian guessed this already in 26 Dec...

On the other hand, 'yes' works. Try both cases against "xclock"
with/without "-iconic". Run it before sawfish / run it and restart 

> Ian> Now I know why it behaves this way.  In events.c, map_request(),
> Ian> add_window() is called first.  That will aplly the matchers, but
> Ian> _then_ the check is made is initial_state == IconicState and if
> Ian> this is true the window is iconified.

It's a mere guess (I don't know window manager's basics, including
WM_**. Nor have I read the code), but the original code is OK,
because the matcher changes the WM_HINTS, so it can precede
initial_state == IconicState.

Second, the patch brings in a new bug, as Timo said:
> XIconifyWindow is inteded for clients, not the window manager.  It
> sends a message to the wm asking it to iconify the window.

(But if Ian was right, then we don't have to revert it. Instead, list
mark the new bug as 'to be fixed'.)

Thank you all, especially Ian, for your enthusiasm. That's what
promote things.

Teika kazura

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