On 08/07/06 18:39, limodou wrote:
> On 8/8/06, Michael Radziej <[EMAIL PROTECTED]> wrote:
>>
>> limodou wrote:
>> > On 8/7/06, Kevin Menard <[EMAIL PROTECTED]> wrote:
>> >> On 8/7/06, Joe <[EMAIL PROTECTED]> wrote:
>> >>> Wouldn't you want to put your database settings (Username and password)
>> >>> in another file as well?
>> >> I would be all for this.  I never liked that the settings file
>> >> contains both project-wide and user settings.  Since it's
>> >> project-wide, it gets added to the SCM.  Since it contains user
>> >> settings, users really shouldn't commit it back.  I realize it's
>> >> Python code and all that, but it seems to be a flaky default to me.
>> >>
>> > So I think secure data should be stored separately from project
>> > settings. Maybe usename and password should be stored in separate file
>> > also. But others don't need to do so. And if username and password
>> > also be stored with secure key together, maybe the appoach what
>> > Michael Radziej  suggested need to be considered.
>>
>> In reality, you will have multiple deployments: production,
>> staging, development, demonstration, depending on your company's
>> policy and what your project is about. For each of these you want
>> to use different databases, might have different middlewares
>> (e.g. for different logging levels, or you
>> might want the CSFR middleware not with your development server),
>> etc.
>>
>> I strongly disbelieve that any fixed scheme of storing some
>> settings separately will cover everybody's needs. It's easy
>> enough to code your own thing in your settings.py.
>>
>> Michael
> 
> So I think different things should be considered different. Just like
> before, many people would say settings.py is only a pure python source
> file, no one would stop you to change it, and make it suit for you.
> That's right. And I think many people like me have done like this. But
> I think a good framework must provide the enough flexibility, but also
> a good way about how to do the right thing. If something are proved to
> be the right and wonderful, why not bring it some spec-like things?
> And if someone don't like this and he still can modify the source
> code, and do what he want. I think if we can introduce such the best
> practice principles in django, it's a good thing.
> 
> And just like many people have said, if you should deploy the product,
> you should think about how to deal with securekey, username and
> password, that's right too. But if there is a default good appoach,
> and it's proved to be good, why we don't use it, and let our head take
> a rest. And I think I can learn many thinks from it.
> 
> Just some thought. It maybe not correct.
> 
> Why I bring this proposal, because I just saw thc settings.py in zyons
> project, and I think it a good way, and I should admit that, I don't
> think of it too much before. And I think there maybe many people don't
> think of it carefully also. So I sumbit this proposal. That's all.
> 

Why not just create a second settings file, call it settings_local.py or 
whatever, and at the end of settings.py do something like:

from settings_local import *

That way you can have settings.py under source control with application 
wide defaults configured and override anything you like in 
settings_local.py for per user or per machine configuration.

You can even have a settings_local.template under CVS with instructions 
to copy it to settings_local.py and change as required.

IMHO a simple, flexible solution for the problem.

--~--~---------~--~----~------------~-------~--~----~
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
-~----------~----~----~----~------~----~------~--~---

Reply via email to