Hi,
I've actually found that these problems are only on the millisecond
time scale, and hence a normal user will never be able to move the
mouse so fast. With recent GUI test changes:
r27444 - http://article.gmane.org/gmane.science.nmr.relax.scm/25195
r27445 - http://article.gmane.org/gmane.science.nmr.relax.scm/25196
the crashing on wxPython 2.8 disappears. This is probably due to the
problem being hidden. However as I don't believe that a user will
ever be quick enough to trigger these issues, I will revert the recent
changes disabling wxPython 2.8:
r27241 - http://article.gmane.org/gmane.science.nmr.relax.scm/24992
r27243 - http://article.gmane.org/gmane.science.nmr.relax.scm/24994
However I will leave the documentation as is, so that users move away
from the buggy wxPython 2.8 version.
Regards,
Edward
On 19 January 2015 at 11:42, Edward d'Auvergne <edward@xxxxxxxxxxxxx> wrote:
Hi,
Because of bug #23189 (https://gna.org/bugs/?23189), I'm considering
depreciating wxPython 2.8 in relax. As it will still be present on
many systems, I am thinking of implementing this as follows:
- The dep_check module gives a warning when the program is started
that the wxPython 2.8 version is buggy and can result in segfaults.
- Disable the GUI tests. The new
General.test_bug_23187_residue_delete_gui GUI test causes the GUI
tests to regularly segfault on wxPython 2.8. The problem is in the
tree GUI element of the spin viewer window. The bug was solved in
wxPython 2.9.
- Updating the dependencies in the relax manual
(http://www.nmr-relax.com/manual/Dependencies.html) and the website
(http://www.nmr-relax.com/download.html#Dependencies, though this is
already done) for wxPython 2.9.
Any suggestions are welcome.
Regards,
Edward
Re: Fatal GUI Segmentation Fault bugs in wxPython 2.8.