[Gsoc] Grok RDB integration (Gsoc student: Rama D. Chavali)

[Philipp von Weitershausen]

On 3 Apr 2008, at 12:22 , Bernd Dorn wrote:
>
> On 03.04.2008, at 10:37, Malthe Borch wrote:
>> On 03/04/2008, Aroldo Souza-Leite <asouzaleite at gmx.de> wrote:
>>> My name is Rama and I attend the University of Houston and am  
>>> majoring in
>>> Information System Technologies.  I have been learning Microsoft  
>>> SQL Server
>>> and Visual Basic for the past year but recently started working  
>>> with open
>>> source software.  I have been introduced to Zope, Plone and Python  
>>> from a
>>> friend and would like to learn more about any of them.  Using SQL  
>>> Server, I
>>> am building database and online management system as class  
>>> project.  I would
>>> like to work on a project for GSoC that would help me further my
>>> understanding of Zope, Plone and Python.
>>
>> You need to define your project more precisely. Zope does
>> RDB-connections and there are numerous high-level packages to make it
>> easy and transaction-safe.
>>
>> That said, I'm sure there's a benefit in having a grokkish approach  
>> to
>> setting it up. Did you have something like sqlalchemy ind mind?
>>
>> Or is this perhaps related to a nozodb-setup?
>
> i don't think that it is a good idea for nozodb to worry about any  
> specific persistence implementation besides the transaction  
> handling, but it could serve as an example for rdb-based persistence.
>
> thinkin twice about it, we have to have a look if we even need the  
> transaction stuff in nozodb

Personally, I find it desireable to use the repoze approach and split  
away as much as possible from zope.app.publication into WSGI  
middlewares. For instance, repoze.tm is a middle-ware that does  
transaction management. If we used that in Zope, zope.app.publication  
would become simpler because the transaction handling could be removed.

repoze has effectively shown that this approach enables a non-ZODB  
object publisher based on Zope 2. I think the same can work in Zope 3.