Developers of Sympa

[Serge Aumont <address@concealed>]

address@concealed wrote:
> This is a nice development.
>
> Will Sympa 6.0 also store the list configuration data inside an RDBMS?
>   
Sympa project is a step by step project and the manpower dedicated to
that project is limited. Yes we will put all configs in database in the
future. We don't known right now if this will be done in version 6.0.
Internal architecture for performances and scalability is one of our
highest priority but we must also tack into account :
- many feature requests and contribs
- easy upgrade processus
- code quality
- documentation and formation
- ...

Serge

> On Tue, Aug 19, 2008 at 02:27:54PM +0200, Serge Aumont wrote:
>   
> > In Sympa 6.0, messages spools will move into database at least partially.
> >
> > sympa.pl prepare the message to distribute and optimized subset of 
> > subscribers. Then it will store messages and recipient subset list on 
> > the  database.  A very simple processus will fetch message and 
> > recipients from database and submit it to the local MTA.
> >
> > With this architecture, it is possible to stop sympa during the 
> > distribution process and later, restart it without duplication of 
> > distributed messages. In addition, message priority is applied after 
> > each submission to the local MTA where in current organization, priority 
> > is evaluated only after a message is completely submitted to SMTP.
> >
> > But the main advantage of using database for spools is that several 
> > processus running on different hosts can do the job.
> >
> > I test it with MySQL and discovered that the default configuration 
> > (/etc/my.cnf) must be changed to support huge messages**| (parameter 
> > max_allowed_packet|**  see 
> > http://dev.mysql.com/doc/refman/6.0/en/packet-too-large.html) .
> >
> > Do you known the limitation for other database supported by Sympa : 
> > oracle, postgres, sybase ?
> >