Experiment from OpenSesame2.9 converted to .osexp (in OpenSesame 3.0) doesn't run completely

grushaprasad

Hi!

I'd created an experiment in OpenSeame 2.9 with 10 blocks (where each block had 120 items) and it ran perfectly. However we updated to OpenSesame 3.0 over the summer and I assumed that the experiment would work as well. However, when I try running the experiment now, for some reason it exists out of the experiment and says "The experiment is finished" somewhere in the third block. So then I ran only the third block and it worked fine. I tried other variations of changing the run if statements so that only specific blocks were running. However it seems like somewhere after running two blocks it seems to exit out of the experiment. It doesn't crash - just exits out of it. When I look at the log files of all the times I ran it, it seems to exit out of it after presenting between 351 and 353 items almost always. (except the one time where it presented only 339 items). I am really at a loss at what could be happening or even where the problem could be coming from.. Does anyone know what might be happening?

Also, would it be helpful if I uploaded the .osexp file even though I can't upload my all the files?

Thank you!
Grusha.

Jarik

Hi Grusha,

It sounds a bit like the issue I reported here:

• https://github.com/smathot/OpenSesame/issues/450

Although I did not have the problem with OpenSesame 3.0.7

But it might be worth the effort to quickly check the number of cycles in your loop items.

Best,
Jarik

sebastiaan

Hi Grusha,

If OpenSesame suddenly and unpredictably stops, it is possible that a hard crash occurred, that is, that Python itself crashed. OpenSesame cannot currently distinguish this from a successful run.

What do you see in the debug window?

Cheers
Sebastiaan

grushaprasad

Hi Sebastiaan!

I see nothing in the debug window! As in, it prints that the experiment has started, but after it ends the experiment, nothing else is printed.. Which is why it was really confusing me even more! Do you have suggestions for adding inline scripts that might help me debug this further?

Thank you!
Grusha.

grushaprasad

Hi Jarik!

I checked it and the number of cycles is 120 (which is equal to the number of items) and it repeats each cycle only once.. So I don't think that might be the problem?

Thanks for the response though!!
Grusha.

Jarik

Hi Grusha,

I guess the followup questions would be:

• What OS are you using (something like Windows 7 64 bit for example)
• What OpenSesame version are u using (something like 3.0.7 for example)
• Could you upload your experiment including all files (optionally using dropbox for example)

Best,
Jarik

grushaprasad

Hi Jarik!

I am extremely sorry for the delay in responding! We are using Windows XP at the lab (because it is set up with Neuroscan and updating might make things complicated..). The OpenSesame version is 3.0.0 and here is the link to the folder! https://drive.google.com/open?id=0B_CdQ3Mn3EVzeV84a2dvUFF2VWM (For some reason I had a really hard time trying to upload it on DropBox as a zip folder.. And Google Drive didn't like zipped files either.. So I ended up putting the whole thing on Google Drive..)

Jarik

Hi Grusha,

After I downloaded your big(!) experiment I noticed a few things that also might be of general interest for others with big experiments:

• You made the wise choice to keep your 3600+ .wav files out of the experiment, but I still saw 4 .wav files in the file pool. I think it's better and clearer to remove them so your .osexp file is script only.
• There were some items in the Unused items 'bin'. I think it's better to completely remove them to keep the experiment as small as possible.
• I noticed you have multiple sequences (secuence1, secuence2, secuence3 etc.) which seem to be the same. If they are indeed the same and only the variables in the loops (block1, block2 etc.) differ, I think it is better to make those loops all run the same sequence (sequence1 for example).
• A comparable thing goes for the blank_initialize sketchpads. You can create one item and right click on it in the Overview area. Then choose Copy (linked). This will allow you to reuse the same item. It also has the benefit that when you would like to change the background color of the blank_initialize sketchpads for example, you will only have to do this in ones in one item.

---

I don't know what the cause of the problem is, which experiment gave you the trouble (grusha_ps_experiment1 or grusha_ps_experiment2) and what version of the parallel_port_trigger plugin you where using.

I do think there is a big change it is a memory problem. You can check this by starting the Windows Task Manager and looking at the memory graph after the experiment stopped. A lot of info on this subject is in the issue I reported here

Best,
Jarik

PS a Neuroscan Synamps also gave us problems. We decided to keep Win XP and take the computer of the network

@sebastiaan this might be another interesting test case for how OpenSesame can get sluggish when experiments get big. Deleting an item from the Overview area takes quite long in OS 3.1.2 for example. The same goes for Permanently delete unused items. I attached the experiment without the 1,3GB of .wav files

sebastiaan

this might be another interesting test case for how OpenSesame can get sluggish when experiments get big. Deleting an item from the Overview area takes quite long in OS 3.1.2 for example. The same goes for Permanently delete unused items. I attached the experiment without the 1,3GB of .wav files

Thanks. Yes, this actually came up in this discussion too. I'm going to see if I can pin down the slowness to something specific, and resolve it.

sebastiaan

I uploaded a prerelease (>= 3.1.3a4) that contains at least optimizations that should speed things up a lot in some cases, such as when emptying the unused-items bin.

• http://files.cogsci.nl/software/opensesame/pre-releases/