bug report: writing postscript files

Post Reply
wlkath
Posts: 18
Joined: Wed Jun 08, 2005 12:39 am
Location: Northwestern University
Contact:

bug report: writing postscript files

Post by wlkath »

Hi,

I'm trying to print a graph to postscript using the Print & File Window Manager. When I select the window and choose "print", the nrniv command window gives errors like:

22 [main] nrniv 568 fork: child 1920 - died waiting for dll loading, errno 11

Any ideas regarding the problem?

Thanks,

- Bill
ted
Site Admin
Posts: 6300
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Post by ted »

Just in case it matters, which versions of NEURON and MSWin are you using?
ted
Site Admin
Posts: 6300
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Post by ted »

Just in case the problem hasn't already gone away by itself--which versions of NEURON
and MSWin are you using?
wlkath
Posts: 18
Joined: Wed Jun 08, 2005 12:39 am
Location: Northwestern University
Contact:

Re: bug report: writing postscript files

Post by wlkath »

It's Windows XP SP2, and the latest version of NEURON -- 6.1.2 (1927) 2007-12-30.

The problem didn't go away by itself, but since it looked like a permission problem, I reinstalled from an administrator account (I don't normally work from an administrator account, only as a poweruser; since the NEURON install script seems to only make desktop shortcuts for the user doing the install, I've previously installed from the account that I use).

The reinstall seems to have resolved this problem. I'm not sure what the original issue was, though. Could it matter whether or not the account used for the install has administrator privileges?

Thanks,

- Bill Kath
ted
Site Admin
Posts: 6300
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: bug report: writing postscript files

Post by ted »

wlkath wrote:Could it matter whether or not the account used for the install has administrator privileges?
Apparently so. Maybe, in printing to PostScript files under MSWin, NEURON has to create some intermediate files in an area that is not writable by ordinary users.

Like you, I use XP (when I use MSWin), run apps as a "poweruser" and not as admin, but also only install as admin after having some problem or other 4 or 5 years ago when I installed an earlier version of NEURON as a non-administrator. Don't recall the symptom, but it was troublesome enough for me to install & uninstall as admin ever after.
wlkath
Posts: 18
Joined: Wed Jun 08, 2005 12:39 am
Location: Northwestern University
Contact:

Re: bug report: writing postscript files

Post by wlkath »

wlkath wrote:It's Windows XP SP2, and the latest version of NEURON -- 6.1.2 (1927) 2007-12-30.

The problem didn't go away by itself, but since it looked like a permission problem, I reinstalled from an administrator account (I don't normally work from an administrator account, only as a poweruser; since the NEURON install script seems to only make desktop shortcuts for the user doing the install, I've previously installed from the account that I use).

The reinstall seems to have resolved this problem. I'm not sure what the original issue was, though. Could it matter whether or not the account used for the install has administrator privileges?

Thanks,

- Bill Kath

I was too hasty. It turns out that reinstalling as administrator didn't really make the problem go away, it just seems to have stopped the error message. I don't get the error message anymore when using the "print and window file manager" anymore, but postscript files created with it are empty. And when I try to generate postscript files using hoc, such as doing shapeplot.printfile("test.ps") or graph.printfile("test.ps"), then I do get the error in the nrniv window again. Postscript files do seem to be generated when using the hoc commands, however.

I seemed to recall from dealing the a previous problem that NEURON calls sed to fix a couple of things in the postscript output; at the very least, there are lines in nrn.def that start *pwm_postscript_filter: sed... I thought therefore, that it might be a problem with the filter. In the past when helping to debug the previous problem I was asked to type

system("printenv")

to see what was going on. Interestingly, this generates the error, too (i.e., [main] nrniv XXXX fork: child XXX died...).

I even get the error if I type

system("ls")

So it seems that the problem is that the "system" command is broken, at least for this version (again, NEURON -- Release 6.1.2 (1927) 2007-12-30) on my machine (an IBM/Lenovo Thinkpad running Windows XP SP2).

- Bill Kath
Last edited by wlkath on Sat Jul 19, 2008 12:12 pm, edited 1 time in total.
wlkath
Posts: 18
Joined: Wed Jun 08, 2005 12:39 am
Location: Northwestern University
Contact:

Re: bug report: writing postscript files

Post by wlkath »

wlkath wrote:
I even get the error if I type

system("ls")

So it seems that the problem is that the "system" command is broken, at least for this version (again, NEURON -- Release 6.1.2 (1927) 2007-12-30) on my machine (an IBM/Lenovo Thinkpad running Windows XP SP2).

- Bill Kath
Here's some additional information:

It turns out that this problem only occurs if one clicks on the hoc file in the windows file manager (i.e., Windows Explorer) in order to run it. If one opens the nrngui first, moves to the proper directory, and then loads the hoc file that everything is OK.

Maybe this will help identify what's causing the problem.

- Bill
wlkath
Posts: 18
Joined: Wed Jun 08, 2005 12:39 am
Location: Northwestern University
Contact:

Re: bug report: writing postscript files

Post by wlkath »

wlkath wrote:
Here's some additional information:

It turns out that this problem only occurs if one clicks on the hoc file in the windows file manager (i.e., Windows Explorer) in order to run it. If one opens the nrngui first, moves to the proper directory, and then loads the hoc file that everything is OK.

Maybe this will help identify what's causing the problem.

- Bill

This problem with the system command doesn't seem to happen for all simulation files -- a file with only the command

load_file("nrngui.hoc")

doesn't cause the problem. There's something in our simulation that seems to be triggering the bug. I've verified this on two separate Windows machines.

If the powers that be are curious to investigate, let me know and I'll send the files that cause it.

- Bill
ted
Site Admin
Posts: 6300
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: bug report: writing postscript files

Post by ted »

I'm not having any problem writing PostScript files at all.
MPC TransPort T3100--laptop with 1.73 GHz Intel Pentium M, 1 GB RAM
WinXP Pro version 2002 SP2
NEURON 6.2.1024 2008-06-13
launched neurondemo
selected Pyramidal
used PFWM to select the Shape plot (window in the lower left corner of the GUI)
PFWM / Print / PostScript
Chose a directory in which I had r/w privileges, and wrote a file called foo.ps
foo.ps size is 53,390, starts with usual preamble (%!PS-Adobe-2.0 etc.) and ends with %%Pages: 0
Acrobat Distiller 6.0.0 5/15/2003 produced foo.pdf whose size is 19,856 and which Acrobat Standard (same version as Distiller) displays nicely.

Do you happen to have cygwin installed on your PC? I mean not just the little bit that is included with the MSWin distribution of NEURON; I mean, as in "I actually installed cygwin itself on this PC."
wlkath
Posts: 18
Joined: Wed Jun 08, 2005 12:39 am
Location: Northwestern University
Contact:

Re: bug report: writing postscript files

Post by wlkath »

ted wrote:I'm not having any problem writing PostScript files at all.
MPC TransPort T3100--laptop with 1.73 GHz Intel Pentium M, 1 GB RAM
WinXP Pro version 2002 SP2
NEURON 6.2.1024 2008-06-13
launched neurondemo
selected Pyramidal
used PFWM to select the Shape plot (window in the lower left corner of the GUI)
PFWM / Print / PostScript
Chose a directory in which I had r/w privileges, and wrote a file called foo.ps
foo.ps size is 53,390, starts with usual preamble (%!PS-Adobe-2.0 etc.) and ends with %%Pages: 0
Acrobat Distiller 6.0.0 5/15/2003 produced foo.pdf whose size is 19,856 and which Acrobat Standard (same version as Distiller) displays nicely.

Do you happen to have cygwin installed on your PC? I mean not just the little bit that is included with the MSWin distribution of NEURON; I mean, as in "I actually installed cygwin itself on this PC."

I have no problem with the above simulation; following the steps you give also produces a postscript file in my case. But note that I don't have any problems if NEURON is run from one of the installed shortcuts; the problem seems to appear only if one runs a hoc file by clicking on it in the windows file manager.

And it doesn't occur for all hoc files, either. I've sent you via a separate e-mail a set of simulation files that seem to produce the problem. Again, it will only happen if you click on the hoc file to run it (after compiling the mod files, of course).

I don't have cygwin installed, but I do use other some programs that are based on cygwin, and which seem to use a limited version of it. But I've checked that this problem occurs on two separate machines, one of them a family machine has only basic software installed, so I think that tends to reduce the chance that it's something on my particular machine that's causing it.

By the way, I note that you are running what must be the most recent development version. I'm running the most recent stable version.

Best regards,

- Bill
ted
Site Admin
Posts: 6300
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: bug report: writing postscript files

Post by ted »

wlkath wrote:the latest version of NEURON -- 6.1.2 (1927) 2007-12-30.
There has been at least one "silent release" of a new MSWin installer for 6.1 since then. If you're still having problems, you might try uninstalling what you are now using, then download & install the very most recent nrn-6.1.setup.exe from http://www.neuron.yale.edu/neuron/install/install.html
Post Reply