Imho the pk field being treated special by bulk_update (as the only
record filter predicate, non updatable) comes from one simple idea -
stable object identity during the updates. For an ORM proper object
identification is rather important to not mix things up, shifting
identities might create havoc further down the road or in internal code,
if it is not prepared to deal with identity shifts. Whether stable
identities are really needed for bulk_update or could be lifted by
making pk updatable I cannot tell for sure.
About record identification by another unique field - imho that should
be safe as long as the unique field can sufficiently identify rows in
question on a table. Implementation might get complicated in multi table
inheritance, where the unqiue field may not reside on the target table
itself (would need a similar tables+rows lookup expansion as done for pk).
Overall this sounds like a valuable API addition to bulk_update, whether
it ultimately could lift the non updatable state of pk itself, remains
uncertain (prolly needs tons of tests and a serious code review of some
internals).
Cheers,
jerch
NB: Btw fetching proper pks upfront from some other unique field is
typically very cheap compared to bulk_update runtime itself, given you
have indexed those columns.
Am 07.06.22 um 02:15 schrieb Ebram Shehata:
I've already created a ticket that ended up with 'WONTFIX' here
<https://code.djangoproject.com/ticket/33770>.
Please read it first.
The scenario that led me trying to add that feature:
Well, I'v a model that has a primary key and unique field...
Every X period of time I've a cron that calls an external API to load
updated data and store it in my Django project db.
Thing is I want to execute the update query without having to have the
pk with me.. I already have the unique field in hand (got it from the
API) and it should be enough to identify what object to be updated..
example code that will do so..
/# call external API and load data as list[dict]/
/objs = list()/
/for record in data:/
/ # Just create instance of MyModel class with values needed../
/ objs.append(MyModel(my_unique_field=data['national_id'], name="New
Name"))/
/# I expect to do:
MyModel.objects.bulk_update(objs, fields=["name"],
unique_field="my_unique_field")/
The current implementation requires every object to have its primary key
set, but, that's not always the case. Also, if the primary key is
my-django-project-generated, there's no way AFAIK to get those PKs
without having to query them from DB, which will be a very expensive
query to run if the data is large.
You can find the POC PR here: https://github.com/django/django/pull/15764
I really thing that'd be a very useful update to add.
I also have a question about this function: We can't use it to update
primary keys.. I'm wondering why ? I want to also add that support to it.
--
You received this message because you are subscribed to the Google
Groups "Django developers (Contributions to Django itself)" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to django-developers+unsubscr...@googlegroups.com
<mailto:django-developers+unsubscr...@googlegroups.com>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-developers/f532b4fd-aba6-44dd-b373-e6e934ff8b67n%40googlegroups.com
<https://groups.google.com/d/msgid/django-developers/f532b4fd-aba6-44dd-b373-e6e934ff8b67n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google Groups "Django
developers (Contributions to Django itself)" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to django-developers+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-developers/3a4655da-8205-48d3-5ae1-477168dcd444%40netzkolchose.de.
