Subject: Re: [nmr_sparky] hanging python anyone?
From: Thomas Goddard
Date: Feb 1, 2006

Previous: 66 Next: 68


Hi Markus,

Does the 99% CPU usage only start after you quit Sparky?

If Sparky is not drawing contours or integrating peaks then its CPU usage
should drop to zero. It does not try to fill idle time with computations.

I am aware of a serious Sparky bug that takes its toll after hours
of use. If you assign many new resonances, the cascaded menu
containing all resonances in the resonance peaks dialog is updated on
each new resonance. Sparky 3.111 does not delete the old menus. This
leads to thousands of undeleted menus (with 200 assigned residues, it
creates 200 new menus each time you add a resonance, ugh). This
eventually crashes Sparky on Windows which can only handle 16,384
menus, but does not crash on other platforms. But it may throw your
X server or the Tk windowing library into a tailspin.

If you dont show the resonance peaks dialog it still gets created
when you click on one of the resonances in the list in the assignment
dialog (another bug -- it wants to update the dialog but should not
be creating it). I believe that is the only time the resonance peaks
dialog gets created where it is not visible. If you avoid that you should
not hit the above problem, though it is mighty inconvenient.

I found this bug last August and have been intending to build a new
Sparky version 3.112 that fixes it but have not done so yet.

This is just a guess at what your problem is. If you are not assigning
new resonances then this is not the problem.

Tom