OK, that is interesting. I can't see why the cursor just being over a dialog should crash the graphics.  Well, I am assuming this is a crash in our C drawing code rather than in Tcl/Tk, or even lower down than that.

I just found this long discussion from 2008/2009 about a bug like this and it turned out to be in a very low level library called xcb:

https://bugzilla.mozilla.org/show_bug.cgi?id=458092

But I would guess that must not be a problem for xcb libraries today (five years later). And I'm not sure when/why xcb would or would not be used by an X11 system.

Wayne

On 7 Aug 2013, at 07:12, Tom Carruthers wrote:

Hi,

and again...

X Error of failed request:  BadIDChoice (invalid resource ID chosen for this connection)
  Major opcode of failed request:  141 (RENDER)
  Minor opcode of failed request:  4 (RenderCreatePicture)
  Resource id in failed request:  0x69f282
  Serial number of failed request:  19067725
  Current serial number in output stream:  19071816
And again - this time I know that my cursor was over the assignment panel, and I didn't click or press anything. It just died by itself.

I did note that the backup file has a time stamp in the OS of exactly the same minute that Analysis died. Also - the back-up (with a time stamp of 15:49) was older than the saved project (time stamp 15:38) - i.e. I think it might have crashed as it was trying to save the back-up file but didn't finish.

I actually remember reporting something like this earlier for Analysis 2.2 (and I have just found it in the archives - February this year "Sudden quit" - not "Sudden quit version 2"). Rereading my initial post shows that the error is the same as the one I reported then. I also noted back then that the Backup had a time stamp of the same time that the crash occurred.

I have tried a memory 'purge' before this last quit - but that doesn't seem to have helped.

Regards,
Tom

On 7/08/13 1:34 PM, Tom Carruthers wrote:
[log in to unmask]" type="cite">
Hi Wayne,

It happened again (the sudden crash). This is obviously a different project to yesterday, but it has the same spectra (but many more as well).

X Error of failed request:  BadIDChoice (invalid resource ID chosen for this connection)
  Major opcode of failed request:  141 (RENDER)
  Minor opcode of failed request:  4 (RenderCreatePicture)
  Resource id in failed request:  0x618fe1
  Serial number of failed request:  20180914
  Current serial number in output stream:  20180976
Once again I was using the assign from roots tool, and everything just closed instantly. It may have happened just as I clicked on the "Assign Root Resonance" button but I am not sure.

This time I had a back-up saved only a few minutes earlier so I didn't loose anything today (see I learned).

Tom

On 6/08/13 6:46 PM, Wayne Boucher wrote:
[log in to unmask]" type="cite"> Oh boy, not good.  The strip creation / deletion seems to be the most prone to crashing, probably because of some timing issue (trying to do some drawing or something before the window is ready or after the window is no longer there). Unfortunately X11 was not written to fail gracefully (unlike Python).

I'd recommend turning auto-backup on (e.g. every half an hour): Project --> Backup.

Wayne

On 6 Aug 2013, at 09:34, Tom Carruthers wrote:

Heya,

I just had a project major crash. Not doing anything different to normal (I don't think) - sudden every window closed in an instant.


X Error of failed request:  BadIDChoice (invalid resource ID chosen for this connection)
  Major opcode of failed request:  141 (RENDER)
  Minor opcode of failed request:  4 (RenderCreatePicture)
  Resource id in failed request:  0x689b66
  Serial number of failed request:  29600307
  Current serial number in output stream:  29601193

I was in process of using the assign from roots tool box and I think it crashed just before I clicked on the 'assign root resonances' button, but I can't be sure. The X11 (XQuartz) didn't close - just Analysis. I don't know how far back in the terminal window the critical error starts - but I remember that I was just working on residue 137Val and that shows up just before the X Error. (As a comment - how hard would it be to time stamp error reports in the console so you can easily see what is relevant to a particular problem?)

I am using an iMac running 10.6.8, with XQuartz 2.7.4, started analysis 2.3.0 with the terminal command "analysis -m 512 proj_file" - that file was closed and a new one was created (I have 20 GB ram available and memory hadn't maxed according to the os)

Arg - and it also turns out I hadn't saved yet, and therefore no backups saved either. A good few hours lost - dumb user error - no programming can fix that!!!

Cheers,
Tom