I got stuck for a little while on a problem with Barnaba while processing a dataset using multiprocessing. Some files caused the program to exit because no valid nucleobase was found. I did not expect this behavior which caused my program to inexplicably stall since sub-processes exited without raising an Exception that I could identify. In my case the problematic piece of code was in the Nucleic constructor. While this is quickly fixable in the caller by explicitly catching SystemExit in a try block, I suggest Barnaba should raise a ValueError to align with python error handling conventions.
Changes
This pattern of using sys.stderr.write + sys.exit is found in many places in the code. I have replaced these occurrences with ValueErrors in the parts that can be used as modules (aka everywhere except in commandline.py)
Motivation
I got stuck for a little while on a problem with Barnaba while processing a dataset using multiprocessing. Some files caused the program to exit because no valid nucleobase was found. I did not expect this behavior which caused my program to inexplicably stall since sub-processes exited without raising an Exception that I could identify. In my case the problematic piece of code was in the
Nucleic
constructor. While this is quickly fixable in the caller by explicitly catchingSystemExit
in atry
block, I suggest Barnaba should raise aValueError
to align with python error handling conventions.Changes
sys.stderr.write
+sys.exit
is found in many places in the code. I have replaced these occurrences withValueError
s in the parts that can be used as modules (aka everywhere except in commandline.py)