Hi Chris,
If you don’t know off the top of your head how to solve this I’ll bug Eyelink, but do you know why it would be that sometimes MWorks fails to connect with Eyelink, or how to fix it? The only fix I’ve found so far is fully rebooting the server computer. Here is the error:
ERROR: Failed to connect to Tracker at 100.1.1.1
ERROR: Error, Eyelink Connection could not be established
ERROR: Can’t start iodevice (Eyelink 1000) and no alt tag specified
Extended information:
parser_context: mw_finalize
Hi Evan,
No, I don’t have any insight into why that would be happening. If you contact EyeLink support, you can tell them that, at the point of failure, all MWorks has done is call set_eyelink_address followed by open_eyelink_connection (which fails).
If EyeLink support proves unhelpful, I can modify the MWorks EyeLink plugin to attempt to get more specific error information from the EyeLink API. (Specifically, it looks like open_eyelink_connection returns an error code that should give a reason for the failure, but the plugin doesn’t report it.)
Chris
Hi Chris,
I’m able to reproduce the error. If I’m in an experiment using Eyelink, and I “close” the experiment before I stop it, the link between the MWorks computer and the Eyelink Host PC never terminates correctly. This seems to prevent future connections until I reboot the MWorks computer. The strange thing is that I also can’t connect to my NI DAQ after this happens. Rebooting the Eyelink Host PC doesn’t fix the problem.
Hi Evan,
If I’m in an experiment using Eyelink, and I “close” the experiment before I stop it, the link between the MWorks computer and the Eyelink Host PC never terminates correctly. This seems to prevent future connections until I reboot the MWorks computer.
That makes sense. MWorks must be failing to clean things up properly in that scenario. I’ll take a look at what’s happening.
The strange thing is that I also can’t connect to my NI DAQ after this happens.
The NI-DAQ drivers are also sensitive about being shut down properly, so I’m not surprised that closing the experiment before stopping it leaves them in a bad state. With USB devices, I’ve found connection failures can be resolved by unplugging and reconnecting the device. Are you using a PCIe card?
Chris
Hi Evan,
To address this issue, I’ve updated MWClient so that the open/close experiment button is disabled while an experiment is running. Hence, to close an experiment, you must stop it first. This change should be in tonight’s nightly build.
Cheers,
Chris