Re: no header columns again

On 24/07/2023 14:49, Duncan wrote:
> dchmelik-Re5JQEeQqe8AvxtiuMwx3w posted on Sat, 15 Jul 2023 18:25:29 -0700
> as excerpted:
>
>> Pan got upgrade to 0.154 on my OS (Slackware64 15+current GNU/Linux) but
>> once again I'm only seeing the last header of subject, author, date,
>> bytes.  Someone said the first three disappeared because of a 'race
>> condition' but I even turned off 'work online', quit, restarted, and it
>> still happened.  How do I get those back?  It's unusable otherwise.
> I see you got them back, but to explain that "race condition"...
>
> Yes, it is a race condition, but work offline won't help, because it occurs
> during GUI setup and has nothing to do with the networking.
>
> The problem has to do with gtk icon-caching.  When they're cached, as they
> are in most cases, no problem.  But when a gtk-related update invalidates
> that cache, the icons aren't fetched correctly the first time (it's not
> waiting for the cache update), so the returned width is zero.  That causes
> the column-width calculations to go screwy.  The result is that anything
> beyond the first icon column (the state and action columns, one of which
> IIRC is the first column by default so it's all of them) gets zeroed out,
> which in turn results in the /last/ column taking up the entire space, with
> everything else squeezed into zero-width columns at the far left.
>
> Then those zero-width column sizes get saved so restarting doesn't help.
>
> The (until next time) user-level fix is to either one-at-a-time drag all
> those stacked zero-width dividers at the far left back to plausible values,
> which works but is fiddly and having to do it over again at the next icon-
> cache invalidation (I've not figured out which package updates the icons
> thus invalidating the cache, but it has to be one or more of the gtk-
> related packages) gets old fast, OR, (manually or automatically) text-edit
> the preferences.xml file to reset non-zero widths.
>
> But editing the prefs file manually gets old after a couple times too, so
> as implied by the parenthetical above, I've automated the process here,
> with a pan startup script that resets them (applying patch files to do so)
> if it finds them set to zero.  So here, when that happens I can immediately
> close and restart pan and the startup script will reset them.
>
> IOW it's not directly a pan bug, but rather gtk related.  Perhaps pan could
> work around it, tho, if someone knowing gtk well enough got hit enough
> times by the bug to trigger them to investigate further.  Unfortunately I'm
> not a dev, just an admin (of my own systems) that reads enough dev-speak to
> have debugged it far enough to have scripted my own mostly-automated
> workaround.
>
> The things we'll do to work around computer bugs in our favorite programs!
> =:^\
>
Not sure about this, as I use windows, and keep building my own pan, but 
shouldn't one run

|gtk-update-icon-cache|

when you update? Maybe that should be in the pan installer?

_______________________________________________
Pan-users mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/pan-users