Add the pulp2_subid field to the Pulp2Content model

[quba42]

fixes #8137
https://pulp.plan.io/issues/8137

The intended use case is to provide one to many mapping from Pulp 2 to
Pulp 3 content. This is achieved by creating multiple Pulp2Content
entries that all share a single pulp2_id, and differ only via the
pulp2_subid.

This is needed for structured Debian migrations:
https://pulp.plan.io/issues/7865

The extra Pulp2Content subrecords are added by the relevant
pre_migrate_content_detail method. The reason is that it is only known
what subrecords are needed once the content is actually inspected by
this method.

[quba42]

Note that the approach was taken from here: #292

The new field is needed for: #273

[quba42]

Should I rename pulp_2to3_migration/app/migrations/0021_auto_20210121_1112.py to something more meaningful?

[quba42]

Not sure why this would cause test failures for RPM.

[ipanova]

Should I rename pulp_2to3_migration/app/migrations/0021_auto_20210121_1112.py to something more meaningful?

yes please.

[goosemania]

Not sure why this would cause test failures for RPM.

Interesting, I wonder if it happens because every null is unique and is a part of uniqueness constraint.
Try to remove null=True and use an empty string as a default.

@dralley , @ipanova , any thoughts on this?

[quba42]

Interesting, I wonder if it happens because every null is unique and is a part of uniqueness constraint.
Try to remove null=True and use an empty string as a default.

I found this in the Django docs:

Avoid using null on string-based fields such as CharField and TextField because empty string values will always be stored as empty strings, not as NULL. If a string-based field has null=True, that means it has two possible values for “no data”: NULL, and the empty string. In most cases, it’s redundant to have two possible values for “no data;” the Django convention is to use the empty string, not NULL.

[goosemania]

@quba42 , your changes look good, thanks!
The CI is temperamental and we are looking at why it is so, it happens to many PRs, not only yours.

[ipanova]

🎉

[quba42]

Question: Should I send something on the mailing list that blank=True is preferable to null=True for string fields? I am pretty sure there are a lot of null=True string fields around the various Pulp repositories. How concerned should we be about this?

[goosemania]

I think it's always good to have such conversations on pulp-dev. I thought we had similar in the past but if we did it was looong ago. +1 to write to the mailing list and see why we didn't go that path in some cases.

[quba42]

Done: https://www.redhat.com/archives/pulp-dev/2021-January/msg00049.html