Re: [Pan-users] Body pane is wrongly resizing pan's top-level window [almost solved]

[Duncan]

walt posted on Tue, 07 Jul 2015 12:37:27 -0700 as excerpted:

> On Tue, 7 Jul 2015 18:46:55 +0000 (UTC)
> Duncan <address@hidden> wrote:
> 
>> walt posted on Tue, 07 Jul 2015 09:46:49 -0700 as excerpted:
> 
>> > Amazing.  The resizing bug I've been describing happens only when I'm
>> > in the tabbed-layout mode (i.e. hit the 'z' hotkey) and, of course,
>> > viewing the body pane.  In non-tabbed mode the resizing problem goes
>> > away like magic.
> 
>> Reproduced!
> 
> Whew, that's a relief!
> 
>> More testing to do, but FWIW, it ignored my normal kwin window rules to
>> do that.  Let's see how strict I have to make them to enforce...
> 
> This detective stuff is fun but time consuming.

Heh, indeed!

What I've found here is that with pan in tabbed mode, the only kwin 
window rule that works is forcing the size.  Forcing max size doesn't, 
and forcing horizontally maximized doesn't.  Those settings are ignored, 
even if the rule also forces ignore geometry requests and do NOT obey 
geometry restrictions.

But forcing size DOES work.

Meanwhile, once size is forced, pan gets a bit more cooperative, and 
rules can be relaxed a bit.  That complicated testing, as sometimes an 
expected resize wouldn't happen, because I'd already forced the size.  I 
didn't test to total reproducibility in this area as there are quite a 
few rule variables involved.

> In the ChangeLog I see this entry:
> 
> 0.94: Ready! Fire! Aim!
> <snippage>
> * fixed resize error in tabbed layout mode. (Robert Marshall, Eric
> Ortega)
> 
> That was done before git, unfortunately.  Can you think of a way I can
> find out what code was patched?  Was there a VCS repo before git?  I
> can't remember back that far :)

IIRC it was an svn repo, hosted by gnome.  Before that I think it was cvs, 
but I think that might have been C pan, before the C++ rewrite, so svn 
likely contains the early C++ history, at least since 0.90 release.  
(Prior to that I think the C++ rewrite project was CKerr's private repo, 
and I'm not sure whether it was initial-lump-checkin or broken down into 
individual commits.)

So the question would be if gnome's SVN repos are still around, 
presumably frozen, for historical reference.

Meanwhile, you could check gnome bugzilla and this list for bugs/posts by 
the listed names.  I have the gmane newsgroup form of the list archived 
here (set unexpiring), and see posts from Robert Marshall dated 
2003-2011.  Eric Ortega has posts from 2002 (early in gmane's history, 
right after it started carrying this list/group and comparable to my own 
first posts here) to 2006.

In fact, Eric's last listed posts are to a "pan 0.93 - resizing" thread, 
in April of 2006.  Robert posts to it as well.  Given the changelog 
mention was for 0.94, that's very likely it.  Actually, the thread starts 
as the 0.93 announcement thread, with Robert changing the subject in a 
followup.

OK, just read the thread and it's apparently the same issue.  Tabbed 
layout problem only, as now, and it was resizing to beyond screen size, 
as now, but in the VERTICAL direction, while we see it now in the 
HORIZONTAL direction.  And identically to here, I couldn't reproduce the 
problem at first, since I was using paned mode and it only happened in 
tabbed.

Hmm... was going to post the gmane web mode post link to the top of that 
subthread, but that was before gmane started adding the header including 
it.  Let's see if I can dig it out of the web interface manually...
This should do it...

http://comments.gmane.org/gmane.comp.gnome.apps.pan.user/5957

So if we can find the related patch, perhaps do what it did vertically, 
horizontally as well?

(BTW, looking at that and remembering, 0.90 was released on April Fools' 
day, 2006-04-01, and that was already 0.93, just 19 days later, on the 
20th.  Pan development has always been fits and starts, and that was 
definitely a going great guns period, effectively weekly releases!  It 
was certainly thrilling, but also prone to developer burnout, which sure 
enough, eventually happened.)

> BTW, I see that I can put pan in non-tabbed mode and then use Ctrl-1 and
> Ctrl-2 to leave only the body pane visible and the bug doesn't trigger,
> so pan is still in non-tabbed mode even though visually pan looks like
> it isn't.
> 
> Hm, no, that's not right.  In tabbed mode pan displays the tabs as well
> as the body pane.  That code could be a place to start looking...

That patch would seem the best place to look if we can find it.  If not, 
yeah, the tabbed vs paned layout, looking for code that treats vertical 
and horizontal differently, possibly with a comment about a bug in the 
vertical.

Meanwhile, as a workaround you can do non-tabbed mode, which doesn't have 
the problem, and use the hotkeys to toggle panes on and off.  (FWIW I 
have a custom hotkey scheme here, so it's not the ctrl-digit keys, it's 
alt-shift-G/H/B for group/headers/body.)

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman