PERFORMANCE. Once started, the serial script runs the processor at 100%. The Infra20 produces 50 8-byte samples each second, while my old computer samples at maybe 10 times that speed. So missing an incoming sample is unlikely. Sometimes, I guess, I get partial samples. If the bytes aren’t all there, or if the sample is all zeros, it gets discarded. I suppose some all-zero samples could be legitimate, but far more often they involve starting the Infra20. The script uses about 1.3 MB of memory, and I haven’t noticed it creeping upward. I buffer 1024 samples then write them out to the hard drive. You can watch the file size grow during the hour, ultimately reaching a size of about 1.5 MB. There is some variation in the # of samples and thus the file size, but normally they are within 1% of each other.

If the serial script has been running for a while new programs sometimes are very slow to start. If I get impatient I stop the script, do my thing, then restart the script.

OPERATIONAL DETAILS. You can unplug and plug the Inra20 in the serial port while the script is running with no ill effects. You can stop and then restart the script whenever you want with no ill effects. You can close the dos window with no ill effects. Obviously this stops the current data capture, but nothing else bad seems to happen. You can run the other scripts (I’ve even done a disk defrag) with seemingly no ill effects (except maybe some lost samples).

Sometimes a sample is corrupted (so far about 1 per million) in the file. The analyzers will complain about it, discard it and continue on.

I’m not sure what happens when daylight savings switches to standard time and back again. I’ll try to remember to look at it when it happens and report back. 2AM – it may not happen… I think the worst case is that it corrupts one or two files at the time of the switch, then continues as usual.