new flush db signal

[michaelritsema@…]

It would be nice to be able to distinguish between a flush db and a syncdb during testing. I use the post_syncdb trigger to load data, it seems odd that I have to do this twice at the beginning of testing. I can't find anyway to distinguish between the two.

[Preston Holmes]

are you testing the args in the signal - it is sent once per app.

Since flush is supposed to mimic syncdb fully - you should not be able to distinguish between them. The emit_post_sync_signal is only called once inside flush - so at this point, in the absence of a clearer description, I'll assume that you are hooking up your test loader to just the syncdb signal, and it is firing for each app (if I'm wrong, reopen with a clearer description of what is happening).

Also note that there is a hook for loading test specific fixtures:

​https://docs.djangoproject.com/en/dev/topics/testing/#fixture-loading

[anonymous]

Django tests uses the create_test_db function. This function does a syncdb followed by a flush. Because I load data based on the post_sync_signal, my data seeding routines are getting called twice. A solution to my problem would be to pass the load_initial_data variable through the post_sync_signal event. This way I can skip my special data seeding routines during this step much like the framework skips its SQL fixture loading.

I could submit a patch myself if you agree.

[michaelritsema@…]

previous comment was from me. I'm reopening.

[Aymeric Augustin]

Note that the behavior of post_syncdb and flush was recently clarified in the docs, see r16874 and r16877.