[ZODB-Dev] Broken instances after refactoring in ZODB

[Laurence Rowe]

Leonardo Santagada wrote:
> On Oct 4, 2008, at 12:36 PM, Wichert Akkerman wrote:
> 
>> Adam wrote:
>>>
>>> Thanks for that, guys, I've not used a mailing list like this  
>>> before so
>>> unsure how to respond.
>>>
>>> If ZODB stores the Package.Module.Class name in the pickle would it  
>>> be
>>> possible for me to simply rename them in the binary file?
> Possible it is, but probably harder than just doing what they said
> 
>>> My confusion here is that I've globally imported everything from the
>>> packages into the current namespace of my main module. ZODB  
>>> shouldn't be
>>> aware I've moved the modules since for all intents and purposes to
>>> Python, they are still there.
>>>
>> It doesn't matter where you import it from or to - python uses the  
>> location of the actual implementation and ZODB uses that. If you  
>> move your implementation to another place you have to either update  
>> all objects in the ZODB or add module aliases.
>>
>> Wichert.
> 
> 
> I would like to know from where does it get that info? I would guess  
> from "__module__".

Correct.

> Why doesn't zodb has a table of some form for this info? I heard that  
> sometimes for very small objects the string containing this  
> information can use up to 30% of the whole space of the file (using  
> FileStorage). How does RelStorage store this?

I believe this was what the python pickle protocol 2 was created for. 
However I think when someone last looked the potential space savings 
with real world data did not justify making the change (Hanno has a 
branch in svn for this).

Laurence