I added exception handling for this in the master branch a while back. Right now, it just adds a warning statement to the logfile and moves on without killing the program. In general, this is ok because these rare failures shouldn't stop what has otherwise been a successful run (chances are in your favor that it wouldn't be an accepted solution anyways since the conversion rate is fairly low). This prevents the loss of a long run and waste of computer resources. However, it can also hide legitimate issues. Maybe it should be more nuanced? Check for why it failed and only continue in certain circumstances? Keep count of how many (or what percentage) of the files are failures and kill the program at a set point?
I added exception handling for this in the master branch a while back. Right now, it just adds a warning statement to the logfile and moves on without killing the program. In general, this is ok because these rare failures shouldn't stop what has otherwise been a successful run (chances are in your favor that it wouldn't be an accepted solution anyways since the conversion rate is fairly low). This prevents the loss of a long run and waste of computer resources. However, it can also hide legitimate issues. Maybe it should be more nuanced? Check for why it failed and only continue in certain circumstances? Keep count of how many (or what percentage) of the files are failures and kill the program at a set point?