[ZODB-Dev] Webkit Threading and ZODB 3.3a2: problems on Windows

[Matt Feifarek]

We've found the culprit. It seems to be a case of pilot error.

With a debugger and slow inspection, I found a tiny bit of code in a 
method that is supposed to be read-only that is setting a small flag 
inside of an object by dipping in and setting an obj._something on the 
sly. That developer will have his pay docked and his milk-n-cookie 
priviledges taken away for a month.

So, this is all elaborate "our bad".

BUT, there is still something smelly. It doesn't seem like changing 
these objects should prevent persistence from happening for the 
remainder of the application's lifetime. It seems that even if the 
developer makes a mistake (like we did) it shouldn't leave the entire 
ZODB behavior unpredictable for other future transactions. Better, it 
seems that we should have some spurious writes, with data we don't want, 
BUT that the "real" writes are reliable. If figuring this out and fixing 
it is a goal of the group, we will certainly contribute as much as we can.

Also, I want to re-iterate that as a beginner (but not a novice) using 
the documentation, it made absolutely no cautions about NEEDING to do a 
commit() or abort(). It's quite clear on the matter of mutables (which 
is good stuff) but not so clear on warning about transaction re-use or 
any of these things you all helped me sort out.

Thanks again to everyone. This is our first dip into the ZODB-dev list, 
and it has been enormously useful.