Open graeme-winter opened 4 months ago
Depending on the environment where it is run (in a test environment to check certain compliances versus within automatically triggered pipelines) a different behaviour might be considers: what constitutes a fatal error on one might not be one in the other.
Within a processing pipeline a user has options to override such issues. This is a crucial mechanism we use in autoPROC: there can be breaking changes in HDF5 datasets inbetween software releases that would stop all processing jobs if they become a fatal error, but either autoPROC recovers automatically (using other defaults etc) or allows users to provide a temporary fixup directly on the command-line.
Anyway, adding an option (--stop-on-error or such) that would trigger your suggested behaviour seems like a good idea, yes.
e.g.
however
imginfo
ran to completion though printed a warning that something was missing: feels like this should have been caught byimginfo
We are happy to look at this also