Re: [Interfacewm-discuss] Re: Some minor patches (was: reminder)

[Chris B. V. <ch...@we...>]

On Tue, 30 Jul 2002 01:17:27 -0400
"e.sammer" <er...@li...> wrote:
> Chris B.Vetter wrote:
> > On Mon, 29 Jul 2002 16:49:00 -0500
> > "Mondragon, Ian" <ian...@ba...> wrote:
> >>could you possibly abuse these & commit them if they work as advertised?
> > Weeeeell, there are two (related) problems I haven't figured out yet.
> > You can see them in action, when/if you start X11 with only a (naked)
> > terminal. That's how I usually go into X, then start the actual window
> > manager by hand(depending on mood, time of day, and constellation of
> > Orion, ...). If you start IWM (having the patches applied as they are)
> > the initial terminal will be mapped - and disappear.
> I have confirmed this (not that it was really in doubt). I'm a bit 
> concerned about this behavior, but a bug is a bug. Honestly, I'm willing 
> to temporarily introduce this bug (see below) to snub the other problems 
> this particular patch solves. (I have commited all patches submitted 
> EXCEPT FOR the iwmwindowmanager.* and iwmscreen.* patches because of 
> this). Ian - what is your take on this?
> As for the problem itself (unmap / map of windows), I ran it past one or 
> two people and came to the following conclusions:
> It gets in the way when:
> 1. You start a non-wm X client prior to the wm (as described by CBV)
> 2. You restart the wm while trying to maintain the other X clients (ala 
> "Restart WM" features found in other wm's)
> Normally, this wouldn't be a big deal because #1 is, arguably, rare. The 
> second problem is a longer term thing and would most certainly be 
> addressed prior to that, so I'm less concerned. The reason it does kind 
> of suck in this case is because #1 is VERY common during testing / 
> debugging / development.

Yes and no. Besides #2, let's assume IWM's signal/error handling gets
expanded to restart itself upon crashing. There may be left over xterms
and apps that would need to be parented again... so this _should_ be
addressed - unless IWM crashing (let's hope it won't) would result in
just a core dump (very unconvenient).

> Due to the fact that a bug would be introduced with these two patches 
> (namely the 2nd hunk in iwmwindowmangaer.m.diff, IIRC), I'm going to 
> hold off on applying said patch until I can talk to Ian.

Well, I wouldn't say it's a new bug being introduced (not that I would
say I don't make mistakes ;-) but I know that it works, having used that
particular code before - there's something b0rken in IWM's mapping or
exposure handling, probably the latter, as Ian already made a bug-report
about it on Sourceforge.

-- 
Chris