pemsley / coot

Software for macromolecular model-building
http://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
GNU General Public License v3.0
122 stars 48 forks source link

Two issues encountered testing Coot 1.1.10 (homebrew) #146

Open olibclarke opened 4 months ago

olibclarke commented 4 months ago

Hi Paul,

Just listing two (possibly unrelated) issues encountered using the latest homebrew build (1.1.10, brewsci-bio-coot) with the tutorial model and data:

  1. Traceback at startup - not sure if this is normal:

    Traceback (most recent call last):
    File "/Users/oc2188/.config/Coot/coot_preferences.py", line 17, in <module>
    coot.show_modelling_toolbar()
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^
    AttributeError: module 'coot' has no attribute 'show_modelling_toolbar'
  2. Ramachandran plot very glitchy/unusable: Here is a screen recording demonstrating what I mean: https://www.dropbox.com/scl/fi/qpvfmjqzuo01minjgf3hn/rama_bug.mov?rlkey=phn14jdm6gwkyuckhxx3rxow6&dl=0

Basically, the Rama plot loads ok, and residues are clickable (yay!). It does affect window resizing (making vertical resizing of the main window impossible), but otherwise it works. However, if I then try to close the rama plot (~0:44 in the screen recording), everything goes bananas - it initially closes, but then randomly reappears, the main window eventually becomes unresponsive and window resizing then causes a black screen. This is on a 14 inch Macbook pro, apple silicon M2-Max, Sonoma 14.5. After killing the unresponsive app, here are a selection of the messages in the log:

(MacCoot:7659): OpenGL-CRITICAL **: 13:56:55.323: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation

(MacCoot:7659): OpenGL-CRITICAL **: 13:56:55.332: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation
(0)  N  /1/chainid="B"/43/VAL, occ: 1 with B-factor: 25.08 element:  N at (144.511,11.001,-15.729)

(MacCoot:7659): OpenGL-CRITICAL **: 13:56:59.597: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:00.289: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:00.298: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:00.661: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:00.764: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation
DEBUG:: --- new_startup_on_glarea_resize() 2106 1364

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:01.663: ../gdk/macos/gdkmacosglcontext.c:202: invalid operation
DEBUG:: --- new_startup_on_glarea_resize() 1970 1364

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:01.674: ../gdk/macos/gdkmacosglcontext.c:202: invalid framebuffer operation
DEBUG:: --- new_startup_on_glarea_resize() 1876 1364

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:01.692: ../gdk/macos/gdkmacosglcontext.c:202: invalid framebuffer operation
DEBUG:: --- new_startup_on_glarea_resize() 1804 1364

(MacCoot:7659): OpenGL-CRITICAL **: 13:57:01.708: ../gdk/macos/gdkmacosglcontext.c:202: invalid framebuffer operation
DEBUG:: --- new_startup_on_glarea_resize() 1696 1364
rbs-sci commented 3 months ago

I get a hard lock on Sonoma 14.5 when opening the Ramachandran plot window. Need to kill Coot. 1.1.10.

Only output is below:

** (MacCoot:41293): WARNING **: 17:06:34.551: Reimplement update_validation(). The function should iterate over the std::map holding validation data for each active graph and recompute it, then trigger a redraw.
update_ramachandran_plot() 0 FIXME?
debug:: accept_moving_atoms:: GTK4 update: update rama plot for 0
ERROR:: in unset_fixed_during_refinement_udd() mol is null

** (MacCoot:41293): WARNING **: 17:06:34.552: Reimplement update_validation(). The function should iterate over the std::map holding validation data for each active graph and recompute it, then trigger a redraw.
update_ramachandran_plot() 0 FIXME?
INFO:: run_post_manipulation_hook_py() No Python error on callable check

(MacCoot:41293): Gdk-CRITICAL **: 17:06:46.109: _gdk_macos_surface_update_fullscreen_state: assertion 'GDK_IS_MACOS_SURFACE (self)' failed

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713c4d0(GtkCellView)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713ab10(GtkBox)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713a990(GtkToggleButton)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713a7e0(GtkBox)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713a660(GtkComboBox)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713a270(GtkGrid)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x12713a0f0(GtkBox)

(MacCoot:41293): Gtk-WARNING **: 17:06:46.115: Broken accounting of active state for widget 0x127138320(GtkDialog)
resize gl rama to 800 800
DEBUG:: --- new_startup_on_glarea_resize() 2674 2264
pemsley commented 3 months ago

ERROR:: in unset_fixed_during_refinement_udd() mol is null

That looks like it might be related.

pemsley commented 3 months ago

Traceback (most recent call last): File "/Users/oc2188/.config/Coot/coot_preferences.py", line 17, in coot.show_modelling_toolbar() ^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: module 'coot' has no attribute 'show_modelling_toolbar'

This was mentioned on the mailing list I think. I have removed much old python toolbar code. We are looking at an old config and a new executable. Hopefully just removing line 17 from your coot_preferences.py will fix the problem.

olibclarke commented 3 months ago

Hi Paul,

I commented out these lines (17 plus the other ones referencing the modeling tool bar). This gets rid of the startup error pertaining specifically to the modeling toolbar, but all the Rama plot glitches persist.

Cheers Oli

rbs-sci commented 3 months ago

ERROR:: in unset_fixed_during_refinement_udd() mol is null

That looks like it might be related.

Happened using the tutorial data, sorry, should have specified.

pemsley commented 3 months ago

https://www.dropbox.com/scl/fi/qpvfmjqzuo01minjgf3hn/rama_bug.mov

Eeek! Poor Coot! I don't treat Coot like that!

Some of that might be GTK4 on macOS. Some is my doing.

OK, I'll see if I can reproduce it.

olibclarke commented 3 months ago

In my hands there is not much required to reproduce - just open coot, load tutorial data, launch rama plot, close it, then try resizing the window - everything glitches and eventually the screen goes black

pemsley commented 3 months ago

OK, I can more or less reproduce this.