[SQLObject] Fwd: [TurboGears] Re: Anyone else fighting `maximum recursion depth exceeded' with SQLObject

[Kevin D. <da...@gm...>]

A solution to the problem I forwarded along earlier: just move the
self._obsolete line up a few. This is a pretty critical bug, because
it's sending people's systems into an icky state.

Kevin

---------- Forwarded message ----------
From: aaditya sood <aa...@gm...>
Date: Dec 28, 2005 12:25 PM
Subject: [TurboGears] Re: Anyone else fighting `maximum recursion
depth exceeded' with SQLObject
To: tur...@go...
Cc: sql...@li...



Kevin Dangoor wrote:
> On 12/27/05, programmer.py <pro...@gm...> wrote:
>
>>The attribute lookup on self._obsolete calls self.__getattr__
>>The first statement in self.__getattr__ is a call to
>>self.assertActive().
>>
>>self.assertActive has one assert statement:
>>assert not self._obsolete, "This transaction has already gone through
>>ROLLBACK; begin another transaction"
>>
>>However, the self._obsolete in the assert calls self.__getattr__ which
>>starts the infinite recursive loop.  I'm not sure of how python
>>resolves attribute lookup.  But from what I can tell, there was no
>>_obsolete attribute (I did a dir(self)) at the time __del__ was called.
>
>
> __getattr__ is called when standard attribute lookup fails. As the
> other poster in this thread mentions, self._obsolete is set in the
> constructor, and I didn't see any hackery going on to get around that,
> and I didn't see anything deleting that attribute, so I'm not sure how
> we're seeing Transactions without _obsolete.

I think I found somethings.

In sqlobject-0.70/dbconnection.py:

class Transaction(object):

    def __init__(self, dbConnection):
        self._dbConnection =3D dbConnection
        self._connection =3D dbConnection.getConnection()
        self._dbConnection._setAutoCommit(self._connection, 0)
        self.cache =3D CacheSet(cache=3DdbConnection.doCache)
        self._obsolete =3D False

I figured that the init fails *before* it has had a chance to set
_obsolete.

I moved self._obsolete =3D False to the first line, and now atleast it
doesn't go into an infinite loop. Dunno if it will screw up some
transactions rollback/commit logic though.

>
> I just forwarded this message on to the sqlobject list to see if
> anyone has any ideas.

Lets see if this helps. And what they say about it.

cheers!
--
/A


--
Kevin Dangoor
Author of the Zesty News RSS newsreader

email: ki...@bl...
company: http://www.BlazingThings.com
blog: http://www.BlueSkyOnMars.com