Dear Paul,

Am 10.02.20 um 16:42 schrieb Paul Emsley:
> On 10/02/2020 11:23, Christian Becke wrote:
>> I am currently testing coot built from the refinement branch (revision
>> 9786) on Gentoo linux.
>
> are you using ccpem jhbuild or using the build-it script? Or something else?

Something else. I wrote an "ebuild" for coot, which is essentially a
wrapper script around autotools used by Gentoo's package manager
"portage".  But I also built it for testing using plain ./configure and
make. I had to patch the configure script a bit to get it to pickup some
of the dependencies (glut, gtkglext, guile and python) - the autoconf
macros that come with coot do not work so well on Gentoo, so I had to
patch in pkg-config support for some of these.

>>  I would like to display maps and models using
>> hardware stereo. Stereo is working fine if I use an older version of
>> coot (e.g. coot-0.8.9.2 EL as shipped with CCP4). Unfortunately, with
>> coot built from the refinement branch, maps are hidden upon switching
>> from mono to hardware stereo. When switching back to mono, maps are
>> shown again. These messages are generated upon switching to hw stereo
>> and back to mono:
>>
>> (stereo-mode-state)
>> (stereo-mode-state)
>> (stereo-mode-state)
>> (stereo-mode-state)
>> (stereo-mode-state)
>> INFO:: Hardware stereo widget opened successfully
>> ERROR:: using display list 0 when main_or_secondary is 1
>
> Oh dear - that look like I left behind some debugging :-(
>
>
> (set-display-lists-for-maps 0) ;; or maybe 1

That was easy! Works like a charm, thank you.

>> There is also something strange about the stereo display of the model -
>> possibly a different stereo angle than in the previous versions?
>
> Yes. First person to mention it (congratulations?). The difference
> between the eyes is now a shear, not a rotation. This means that you 
> don't get part of the structure and map that is in one eye but not the
> other - which should improve (as in reduce) eye-strain. There are still
> weird visual effects if you letterbox the window.

Good to know, I will test if it works for me.

>> Another, probably unrelated, issue I am facing are random crashes during
>> startup. Messages look like scheme plugins try to access GUI code that
>> has not been initialized, yet (see below). If I restart coot several
>> times, it will eventually startup successfully after a couple of tries,
>> so I suspect some kind of race condition.
>
> This doesn't ring any bells.  Can you use gdb on the core file?
I don't have a core file around right now. Will try to get a backtrace
next time it happens.
>>  I applied some patches to
>> build coot against guile-2.0 (guile-1.8 is not available on Gentoo
>> anymore), maybe this is causing problems? I would appreciate some help
>> with debugging this.
>> load /home/chris/.coot-preferences/curlew-morph-march.scm
>> Backtrace:
>> In ice-9/boot-9.scm:
>>  160: 4 [catch #t #<catch-closure 7f8f8d98c820> ...]
>> In unknown file:
>>    ?: 3 [apply-smob/1 #<catch-closure 7f8f8d98c820>]
>> In ice-9/eval.scm:
>>  411: 2 [eval # ()]
>> In coot-gui.scm:
>> 1054: 1 [coot-menubar-menu "Morph"]
>> In unknown file:
>>    ?: 0 [gtk-container-children (#{score-1 1459}# . #f)]
>>
>> ERROR: In procedure gtk-container-children:
>> ERROR: In procedure gtk-container-children: Wrong type argument in
>> position 1: (#{score-1 1459}# . #f)
>
>
> Yes, this looks like the sort of thing to expect if one is using the
> wrong/new/modern/supported version of guile-gtk - it looks like
OK, so I am shooting my own foot, here - I was kind of expecting that. :)
> coot-main-menubar isn't returning a value that coot-menubar-menu can
> deal with - I will need to look at the glue between guile and c/c++.

What I still find puzzling is that it does not always crash loading the
same script, last lines of output before the crash are not always the
same. And it doesn't always crash. So sometimes coot-main-menubar is
returning the right value, sometimes not? Still some kind of race
condition during initialisation?

> If needed, you can compile coot without guile and it works pretty good.

That would mean I would have to live without many of the curlew
scripts... :(


Thank you for your help and fast reply,

Christian

########################################################################

To unsubscribe from the COOT list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=COOT&A=1