On Wed, Oct 21, 2009 at 2:55 PM, Hanne Moa <[email protected]> wrote:
>
> On Tue, Oct 20, 2009 at 16:18, Russell Keith-Magee
> <[email protected]> wrote:
>>
>> On Mon, Oct 19, 2009 at 11:52 PM, Hanne Moa <[email protected]> wrote:
>>>
>>> It doesn't look like Contrib-06, covering ticket #3011, Allow for
>>> extendable user module, will be in for 1.2. This is
>>> my pony for *2.0*:
>>
>> I completely agree that we need to do something about #3011. /../
>> However, the time has really past for big ticket discussions for v1.2,
>> and it certainly isn't the time to talk about ponies for v2.0.
>
> Saying it's for 2.0 makes it possible to ignore backwards compatibility...

Possible doesn't necessarily mean encouraged, and there still needs to
be a clear migration path.

On top of that, 2.0 isn't "just around the corner" - its several years
away from even being a likely topic for discussion.

Lastly, if we can find a deprecation path rather than a backwards
incompatibility, we don't have to wait until 2.0 - we can do it in a
series of v1.X releases.

>> One of the big problems that needs to be solved that you haven't
>> addressed is the migration path. Obviously, we can't just replace the
>> current auth.User model - we need to have a migration plan for all
>> existing users of contrib.auth.
>
> auth.User as an "auth-profile" on the MinimalUser, then it could start
> out looking exactly as today except having its primary key be a
> foreign key. Then we could make some of the other fields on auth.User
> fetch its data from minimalUser.

That is an interesting approach - I hadn't considered that. There
might be some complications around models that have existing foreign
keys on User - obviously, they will continue to work, but if a foreign
key on MinimalUser is more appropriate, there is a different migration
issue.

> The tricky part IMO is getting 3rd parties to use the MinimalUser so I
> foresee the big job as to rewrite tons of examples, tutorials, howtos.
> Pushing it, basically. Oh and there should be screencasts, I love
> those.

... And I want a pony :-)

>> I have some ideas on how this could be done, but I'd be interested
>> in hearing what others come up with. There are also complications
>> with integrating with admin that need to be worked through.
>
> Sometimes it seems babying the admin is keeping us from doing a lot of
> cool stuff. I couldn't use the admin at all in my very first django
> project (legacy database + cranky DBA), so I guess I never grew used
> to having it available.

Like it or not, the admin has always been a big draw card feature of
Django. I know that the admin isn't perfect for every situation, but
I've also lot track of the amount of time it has saved me over the
years. I can guarantee you that any proposal that starts with "lets
ignore the admin" isn't going to get much traction with the core team.

Yours,
Russ Magee %-)

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"Django developers" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/django-developers?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to