That made us aware of the fact, that the database wrapper for PostGIS
execute a SQL statement every time it prepares the database.
That behavior seems inconsistent with how any other PG extension works.
Usually, those have to be initialized in a migration explicitly. There has
been a debate in the past, if that statement can do any harm, but clearly
Heroku has proven it can. Therefore, I would suggest deprecating this
behavior and schedule it for deletion in Django 5.
Superseeds #28975
--
Ticket URL: <https://code.djangoproject.com/ticket/33885>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.
* keywords: fly.a...@gmail.com =>
--
Ticket URL: <https://code.djangoproject.com/ticket/33885#comment:1>
* cc: amureki (added)
--
Ticket URL: <https://code.djangoproject.com/ticket/33885#comment:2>
* status: new => closed
* resolution: => duplicate
* easy: 1 => 0
Comment:
Thanks for the reports, It's definitely a duplicate of #28975 (that we can
reopen, if needed).
--
Ticket URL: <https://code.djangoproject.com/ticket/33885#comment:3>
Comment (by Johannes Maron):
Replying to [comment:3 Mariusz Felisiak]:
> Thanks for the reports, It's definitely a duplicate of #28975 (that we
can reopen, if needed).
Thanks for the swift triage. I didn't feel comfortable to reopen it, since
it was about skipping it in Python and I would rather not make the call at
all. But I can reopen it and make my case without changing the ticket
description.
--
Ticket URL: <https://code.djangoproject.com/ticket/33885#comment:4>