syncdb fails if the loaddata management command is called in a post_syncdb handler

[Keith Bussell]

Calling management.call_command('loaddata', ...) from within a post_syncdb handler will cause all subsequent SQL to fail. This broke as of r7803.

models.py:

class Book(models.Model):
    title = models.CharField(max_length=255, db_index=True)

management.py:

def post_syncdb(sender, created_models):
    if models.Book in created_models:    
        management.call_command('loaddata', 'data.json')
dispatcher.connect(post_syncdb, sender=models, signal=signals.post_syncdb)

data.json:

[{"pk": 1, "model": "app1.book", "fields": {"title": "This is a test"}}]

> manage.py syncdb
Creating table app1_book
Installing json fixture 'data' from 'C:\source\test\app1\fixtures'.
Installed 1 object(s) from 1 fixture(s)
Installing index for app1.Book model
Failed to install index for app1.Book model: cursor already closed

[Keith Bussell]

Forgot to mention, in case it's relevant: using the postgresql_psycopg2 engine against PostgreSQL 8.3.

[Keith Bussell]

Added a simple patch to fix this.

[Keith Bussell]

actual fix

[Russell Keith-Magee]

(In [7889]) Fixed #7597 -- Added code to force the re-opening of the cursor connection, just in case a post-syncdb handler closes it. Thanks to keithb for the report and fix.