bulk_create() does NOT send signals

[YassouSr]

I am using bulk_create() function to insert multiple records, however all changes made by this function are not logged correctly with CRUD event :

            with transaction.atomic():  # Ensures atomicity
                created = Inscription.objects.bulk_create(valid_inscriptions)

            # throw save signal for log easyaudit
            # since bulk_create bypass post_save signal
            for inst in created:
                post_save.send(sender=Inscription, instance=inst, created=True, raw=False)

I have tried the above solution to call signal but it throws error : TypeError: post_save() missing 2 required positional arguments: 'using' and 'update_fields'

I also searched that it is not a good idea to call save signal manually, and also due to large data I cannot loop through data and save records one by one due to optimization problems and response time.

Any other solution would be appreciated.

[jheld]

It's been many years since I tried to construct a system for this.

My high-level analysis is that the library could provide support functions/context managers or something to facilitate the flow. I do think it would useful to provide it. However, as you've noticed, there is no current way to leverage signals for bulk operations (and even if there was, not all databases support returning IDs) and thus the library cannot hook to them.

But the more I think about it, we cannot provide it for non-postgres. Even if we wanted to pass in instances from the model like model(...) as opposed to model.objects.create(...), and eventually submit those for bulk create, we still would have no direct link between a given object and what row was actually created. We cannot do it, unless we build support explicitly for postgres - which is something we could choose to do. In that case, we would already have the objects in memory and we could simply attach the ID and then pass them all into the library API.

If there's something I'm not aware of, please let me know.