Thomas Goirand <> writes:

> Feel free to raise this on a Django upstream message. I very much agree
> this isn't best.

I believe there is (or at least was, and probably still is) general
agreement that the Django settings mechanism is horrible for numerous
reasons. IIRC this has been discussed at past PyCON AU conferences.

Unfortunately, this has been labelled as a "hard problem". As in coming
up with a replacement that the Django core developers can agree to is
non-trivial (which also means dealing with existing applications in a
sensible manner).

e.g. do a google search for "django settings class" and you will see
that there are a number of projects to get class based settings for
Django. A feature I really wish Django supported out of the
box. Unfortunately most of these projects (I might have missed
something) appear to have been abandoned.

Maybe this is an isssue that needs to be pushed at PyCon or PyConAU.

If there isn't an existing upstream bug report already, somebody should
definitely file one. Or add details as to how the problem affects
testing, if not already given. I don't believe the bug report by itself
will solve this problem, but I think it is a necessarily prerequisite
for a solution.
Brian May <>

Reply via email to