Saving data in case of crash?

[David McFarlane]

Looking at the DirectRT output data files, I see that trial data are saved in the same order as in the input file. I don't see how DirectRT can do this unless it keeps all results in memory and then writes them all to disk at the end of the session. Is that the case? If so, then what happens if the system crashes in the middle of a session? Do I just lose all my data?

Thanks,
-- David McFarlane, Professional Faultfinder

[jarvis24]

DirectRT does wait until the end of the session to write the data. When the code was written for DirectRT, it was a high priority to avoid any system activity possible during the session that could interfere with the precise timing of events and response times. We included data file I/O in this category. And so yes, in the event of a system crash during a session, DirectRT would not be able to create the data files for the data gathered to that point.

[David McFarlane]

Thanks for the direct & frank answer. But doesn't DirectRT do disk I/O to read stimulus files during the ITI in order to prepare the next trial, without affecting performance during the trial? If so, couldn't DirectRT just as well dump some data during the ITI? As you indicate, it comes down to a tradeoff between performance and data safety, I just wonder if every customer wants to make the same tradeoff.

-- David McFarlane, Professional Faultfinder

[jarvis24]

Fair point. I guess what would be ideal is to make it optional. At the end by default (to maintain backward compatibility) and have a optional check to write it as you go.