Closed danielrichman closed 13 years ago
This is a great idea, will try and get round to this at some point!
I was looking at predict.py on nessie; I think that if you have the python script daemonise itself by fork disassociate fork, then you can avoid using atd, its forked process and the extra sh
process that it will use.
Done.
Feature request/suggestion; maybe calling signal.alarm() at the top of the script with a large value would present the build up of stalled predictor processes.