Bug report OS 0.27~pre17

Daniel

Platform: Win XP/7

Bug: When drawing a gabor patch with orientation 0, 90, 180 or 270 degrees (e.g. a horizontal or a vertical orientation), only the mask seems to be drawn. You just see a white orb whose edges fade into the background: the black and white frequency pattern that should be drawn inside it does not appear.

---

Michif

Just in addition to that a FEATURE SUGGESTION:

Would it be possible to implement option settings to change the format of the data output?

So far the only way data is saved is in rows and only a single colomn. Although I know there are work-arounds in excel, I would really appreciate a function to change the style of the output. This could be for instance "every variable in a new colomn" or even wrap-ups from several trials such as "average every 10 trials". Especially the first, could improve data import in other programmes such as MATLAB or SPSS.

sebastiaan

Bug:When drawing a gabor patch with orientation 0, 90, 180 of 270 degrees (e.g. a horizontal or a vertical orientation), only the mask seems to be drawn. You just see a white orb whose edges fade into the background: the black and white frequency pattern that should be drawn inside it does not appear.

For me the Gabor patches look fine. Could you perhaps provide some more details (backend, specific parameters of the Gabor. etc.) and a screenshot?

So far the only way data is saved is in rows and only a single colomn.

I think something went wrong here, because the data should be formatted as you suggest: One trial per row (or actually one logger call per row, which is usually one trial) and one variable per column. If the logfile looks funny, perhaps you can try opening it in a different program? Or otherwise you can open a new discussion and share the logfile using PasteBin so that someone can help you find out what went wrong.

Cheers!

Daniel

For me the Gabor patches look fine. Could you perhaps provide some more details (backend, specific parameters of the Gabor. etc.) and a screenshot?

The following code (for simplicity all done in run phase, but results are the same when drawing in prepare phase) produces the screenshot below it:

cnvs = self.offline_canvas()
xc = cnvs.xcenter()
yc = cnvs.ycenter()
cnvs.gabor(xc-100,yc,0,40)
cnvs.gabor(xc,yc,25,40)
cnvs.gabor(xc+100,yc,90,40)
cnvs.show()
self.sleep(3000)

All back-ends show this same result (although with psychopy the gabors' backgrounds are black):

Thus, all orientation values that should either produce a perfectly horizontally or vertically oriented gabor patch, produce a 'blank' gaussian distribution.

Michif

Hey sebastiaan,
thanks for the fast response. We figured it out. It has rather something to do with the language settings. If list seperator is not a comma in Windows, Excel does not transform the stuff automatically in different colomns. Don't know whether there exists a work-around that is bullet-proof (for any language settings in Windows).

-M

Daniel

So far the only way data is saved is in rows and only a single colomn.

This is because Excel doesn't interpret the list separator correctly. In US syntax, this is a comma, but when the regional settings are set to Dutch, Excel expects list values to be separated by a ";". If you open a csv file in which the fields are comma delimited with Dutch regional settings, Excel simply puts the whole string of values in one single column. You can solve this either by changing the list separator symbol in your windows regional settings from ; to , or you can select the complete single column in Excel (>2007), then go to the "Data" tab and choose "Text to column". Indicate "," as your list separator there

sebastiaan

All back-ends show this same result (although with psychopy the gabors' backgrounds are black):

Right! The reason for this is a bit subtle. The frequency is in cycles/px, so it should be a value below one. Basically, you have set the frequency of the Gabor so high that it's not possible to be displayed correctly. Because of anti-aliasing, you nevertheless get an artefactual low frequency texture. But whether this happens depends on the orientation of the Gabor. You can get other funky effects this way as well.

But the bottom line is: Stick to frequencies in the <1 range!

Daniel

Ok, solved! thx

Michif

Feature request:
I would really appreciate a function in the code editors of e.g. an inline script that allows me to mark a whole bunch of lines and comment or uncomment them with one click just like in MATLAB. It could work just like the "indent one level up" function. Such a function is quite handy when trying some stuff out.

Edit:

Maybe also with a short-cut/hotkey for easy and fast access. That would be even more awesome

Daniel

I second the above! It would be handy to be able to (un)comment complete blocks of code at once. Is it possible for you to easily implement this?

sebastiaan

That's a good idea, and should be fairly easy to implement: https://github.com/smathot/OpenSesame/issues/107

It will have to wait for 0.28 though, as I want to focus on getting the last bugs out of 0.27 (and new features may lead to new bugs, etc.)

Cheers!

Michif

Bug report:

When I use the variable wizard in a loop, fill in the first row for the variable name, delete it afterwards and then confirm my input, the input does not appear as a list as usual but stays blank. If I then switch from one element to another and back to the loop I just modified, the list is still empty. However, in the information line it says at the end (behind e.g. "random order") (zero or negative length). After a little bit of bugging it mostly works after the second time I retyped the stuff in the wizard.

sebastiaan

Thanks for the report, that's exactly the kind of polish that we need! Fixed on GitHub, and will be included as of 0.27~pre19.

Michif

Feature suggestion:

I am really looking for a function to merges several experiments into one. I prefer using a seperate file for each subpart of a big experiment. But eventually all of them have to be merged into a single file/experiment that contains all subparts.

So far, I have not found a function that is able to do that (is there?). That means if I want to run all subparts in one I need to copy paste element by element into a masterfile which is really time consuming!

I saw that there is this blue OpenSesame item at the top of each new experiment (in each template). I do not know the underlying structure but wouldn't it be possible to write function that copies the contents of the blue OpenSesame icon (which is probably always the whole experiment) in another file. By that merging several experiments (blue OpenSesame icon and their contents) should be fairly easy, is it?



Minor bug report:

I have run my experiment several times with the legacy engine and the option "run experiments in a seperate process" turned on. The weird thing that happens now is that everytime I have a e.g. 600ms timeout in my keyboard responses is that the timing will be saved as 601ms (correspondingly after 10 trials of me not reaction the average response time is 601ms instead of 600). Although PsychoPy is better for temporal precision there is no variation in the timings aka. its precise (which would be of course not the case if I would run it not as a seperate process). It is a really minor flaw yet confusing. I haven't checked whether this is also the case for other running engines (not even sure whether this matters if I run it as seperate process).

Thanks! Again: Keep up the great work!


Edit: Bug report:

Just discovered this maybe it is not a bug but it does not appear to be logical. When I add a sequence item from the commonly used panel (default on the left side) and add it to the blue OpenSesame item, any other sequence just gets delete. So in fact I am not allowed to have more than one sequence under the blue OpenSesame icon?

sebastiaan

So far, I have not found a function that is able to do that (is there?). That means if I want to run all subparts in one I need to copy paste element by element into a masterfile which is really time consuming!

What you could do is simply open your .opensesame files in a text editor (or via the built-in script editor in the general tab) and merge them this way. Basically, you should be able to copy-paste all scripts into a single file, save it, and open it in OpenSesame. You will need to restore some of the structure, of course, but it should be possible.

The weird thing that happens now is that everytime I have a e.g. 600ms timeout in my keyboard responses is that the timing will be saved as 601ms

Right, thanks for pointing that out. The timeout is now interpreted as the highest allowed response time, so that a timeout value of 1000ms gives timed out response times of 1001ms. But it would be more intuitive if the timeout is interpreted as the lowest not-allowed response time. I'll change that.

When I add a sequence item from the commonly used panel (default on the left side) and add it to the blue OpenSesame item, any other sequence just gets delete.

What you do in this case is replacing the entry point of the experiment (the old entry point will still be under 'unused items', so it's not deleted). In a sense it's valid behaviour, but it's not very intuitive. In 0.27 you should receive an error message when you try this, something like "Cannot change experiment entry point".

Cheers!