Re: [SQLObject] Inheritance looses column

[Oleg B. <ph...@ph...>]

[to the list...]

On Mon, Sep 07, 2009 at 01:34:51PM +0200, Tobias Weber wrote:
> On 07.09.2009, at 11:15, Oleg Broytmann wrote:
>
>> I don't think it's inconsistent.
>
> Well, getting a parent's column sometimes works and sometimes doesn't  

   Yes, you have a point.

> (depending on the cache?)

   Depending on if the entire tree of inheritable objects has been fetched.
Or if you access an attribute that was or was not fetched.

>> Any property is free to raise any exception.
>
> During normal operation descriptors are only supposed to raise  
> AttributeError (which this one does). Whether reading a field at that  
> point constitutes an allowed operation depends on the documentation of  
> _init. I didn't find much, but then it's a bit hard to search for ;)

   :(

>> Looks like a bug in that library.
>
> http://bugs.python.org/issue1785
>
> This solves my problem!

   Aha, that's better!

> I'm still not sure if SQLObject is behaving correctly, though. What's  
> the point of _init if you can't access data yet?

   In this particular case _init is called in the middle of fetching
children. InheritableIteration.next() calls obj.get() which in turn calls
parent.get(), etc; every .get() calls _init, but to access all attributes
you have to wait until all .get's are finished.

Oleg.
-- 
     Oleg Broytmann            http://phd.pp.ru/            ph...@ph...
           Programmers don't die, they just GOSUB without RETURN.