Closed LogosOfJ closed 7 years ago
Thanks again, @LogosOfJ for your PR! The argparse.Namespace
problem you found could also bite us in other code locations. Are you aware of any? And have you tried the code on Windows7 as well?
Thank you @LogosOfJ, not only for the fix (this was a real bug you discovered), but also for making me better understand argparse
concepts.
Thanks, @ulif, and sorry for the delay. What do you have in mind for regression tests?
Thank you, @LogosOfJ, I guess the tests you fixed are in fact regression tests, because they were faulty before and would have failed with all prior versions. I am therefore completely happy with your fix.
In the beginning I thought this was a Windows-specific thing, but it was not. In fact the realdice
source was broken for quite a time and on all python version :-/
I am really glad you found that bug and will now try to investigate how and when I introduced it.
Then, a new version has to be released, but this is my job. Thanks a lot!
In Python 2.7 and 3.5 on Windows 10, using the 'realdice' randomness source fails "out of the box" because the constructor for RealDiceRandomSource unsuccessfully attempts to access an argparse.Namespace object as a dict. This change fixes that issue and updates a test for the constructor.