[open] On Forms and Keyboard Entry Issues...

aberen03 · March 2013

Hello,

So I built my experiment around the use of one looped inline-script item, which references different values in the loop table. Importantly, this inline-script implements a form with 7 items: 1) text display that varies, (2) a picture that varies, (3) a text display with a question, (4) a text_input box, (5) another question, (6) a rating_scale with 7 items, and (7) a next button. Everything works as it should, and it was surprisingly easy to implement (way to go OS team!) - but this is only the case if I use one of my personal computers.

Here's what I mean:
My experiment runs perfectly fine on any of my up-to-date computers (school-issued dell laptop, personal macbook, and personal custom-built PC); but when I did a test run of the experiment using the university computers and the portable distribution, I noticed a problematic latency during text entry. That is, there is a latency between a keystroke and the text actually showing up in the form 'text_input' component, and I can't have subjects participate in the experiment if the program runs that way. I fear it is due to the older and less capable components inherent of university machines.

To remedy, I tried removing a couple form elements, per the recommendations on the "form troubleshooting and performance issues" page in the documentation, and this seems to help some, but the latency was to still noticeable. In addition, removing those elements loses some information that I would like to provide. Is there another way that I could get the same functionality as a form without the slowdown? For instance, is there a way to implement the text input and rating_scale items in a sketchbook or canvas? Can these items only be implemented in a form, or would I likely experience similar text entry latency with any script due to the number of items?

I would really appreciate any advice or recommendations you can give!

Thanks,
Alex

p.s. - I should note here that this is essentially just a survey, where subjects will identify an image and rate it for relatedness to a category. So, timing is not really crucial to the program, but the text input needs to work well. If anyone has an option that is better suited to this kind of thing - like a free web survey - I think OpenSesame might be overkill. I basically just need to display a word along with an image, then a space for text entry and a rating scale. Thanks!

sebastiaan · March 2013

Hi Alex,

Did you try switching to a different back-end and changing the back-end settings, as described here?

http://osdoc.cogsci.nl/forms/performance-issues-and-troubleshooting/

It's true that forms can be really slow, but given the optimal back-end configuration I haven't yet experienced any problematic latencies. But then again, perhaps the systems are really slow, in which case using forms, which are inherently inefficient, might simply not work well enough.

Alternatively, you could try the Expyriment TextInput object, which is much more efficient:

http://documentation.expyriment.googlecode.com/hg/api/expyriment.io.TextInput.html

from expyriment.io import TextInput
ti = TextInput("My question goes here")
response = ti.get()

It doesn't look as nice though. PsychoPy also comes with GUI capabilities. However, these pop up in a different window, which is also suboptimal.

http://www.psychopy.org/api/gui.html

from psychopy import gui
info = {'Observer':'jwp', 'GratingOri':45, 'ExpVersion': 1.1, 'Group': ['Test', 'Control']}
infoDlg = gui.DlgFromDict(dictionary=info, title='TestExperiment', fixed=['ExpVersion'])
if infoDlg.OK:
    print info
else: print 'User Cancelled'

Hope this helps!

Cheers,
Sebastiaan

aberen03 · March 2013

Hello Sebastiaan,

Thanks for your reply.

I have tried the various back-ends, and even using legacy without double-buffering, the text entry latency is problematic.

I tried the expyriment text_input method, but I was unable to devise a way to also present an image simultaneously. Learning to use psychopy is still on my to-do list for the long term, but for now I can't quite wrap my head around it.

For now, I think my best bet is to try splitting up the various form items to separate forms, to reduce the processing burden of each form. As a backup, I'm also looking for a way to implement my survey on regular old survey site.

As I said previously, this is for a pilot study within which timing is not important, but since it's essentially a survey, text-input absolutely needs to be fluid. My plan is still to use OpenSesame for an RSVP study that will follow the pilot, but I'm beginning to fear that my abilities are too limited to make use of OS for this simple pilot.

Lastly, none of this would have been a problem if I could run subjects through the experiment on a one of my personal machines, but, alas, the school computers are dreadfully underpowered :P

Thanks for your help, and let me know if you have any other ideas!

-Alex

aberen03 · March 2013

Me again!

I just tested the approach outlined above - splitting the form elements among multiple screens, and the latency is gone. The only hurdle left is to see if this improvement carries over to the school computers I need to use for the experiment.

I'd still be very happy to hear any advice or suggestions you have

-Alex

sebastiaan · April 2013

Hi Alex,

Ah, I initially missed that you had multiple text inputs on one screen. I can see how that can be really slow.

I'm afraid I don't have any better solution than splitting the form across multiple screens, as you did. The reason that forms are inefficient is that they have to work on all back-ends, and thus cannot make use of back-end-specific hacks to improve performance, such as refreshing only part of the screen. (In the future I hope to find a way to improve this.)

Cheers,
Sebastiaan

aberen03 · April 2013

Hi Sebastiaan,

The design tweak helped performance immensely, and the program now works great on the school computers as well. I hope that some kind of 'performance hack' can be done with forms in the future, they make it quite convenient to present multiple types of content at once. However, as far as compromises go, I think splitting the content among multiple screens really won't affect my study.

Thanks for all your help,
Alex

aberen03 · April 2013

I'm in the process of programming another experiment, and analyzing data from my previous experiment. It has truly been a pleasure using all your software. I have never had an easier or more rewarding experience in programming an experiment. I have been actively encouraging other members of my department to look to OpenSesame and Python for their programming and data analysis needs. I recently ordered a Nexus 7 (poor timing with the update coming, but oh well!) and a Raspberry Pi to show off OpenSesame to the faculty if I get the chance. Keep up the good work.

sebastiaan · April 2013

Thank you, and good luck with your studies!

andersony3k · October 2013

I tried to use widgets to make a form that has two text input fields on one page. I'm using the legacy backend to avoid latency issues. What I've found is that there are some bugs. One first has to enter text into the second input field. Only then can one use the mouse to activate the first field and enter text there.

sebastiaan · October 2013

What I've found is that there are some bugs. One first has to enter text into the second input field. Only then can one use the mouse to activate the first field and enter text there.

Could you share the script that triggers this issue, please?

Cheers,
Sebastiaan

Howdy, Stranger!

Categories

[open] On Forms and Keyboard Entry Issues...

Comments

Howdy, Stranger!

Quick Links

Categories

[open] On Forms and Keyboard Entry Issues...

Comments