Closed arnaudon closed 3 years ago
do we know where it complains exactly?
Isn't that line 'protected' by skip_repair_examplars?
Ok, but so that's a different case, i.e. it checks the 'unrepaired' morph. Imo this morph should always be available, so that we can test a run with the original morphology. The skip_repair_examplars is just there in case the unrepaired examplar is not sure to survive the repair process (which is not guaranteed).
Ah yes, you are right. Indeed, the issue was that @alex4200 tests had missing exemplars, so it was crashing here. I agree we should have these morphs available, but I naively thought that this option would also not require them. All cleared then! Thanks!
@alex4200 noticed that if we don't have the exemplar morphs in the morph release MM complains even with skip_repair_examplars:
https://github.com/BlueBrain/BluePyMM/blob/master/bluepymm/prepare_combos/create_mm_sqlite.py#L118
maybe we can move this if further up to not even try to load repaired exemplar?