SQL backend consolidation

[Mark Kretschmann]

On 8/16/07, Seb Ruiz <ruiz at kde.org> wrote:
> On 16/08/07, Ian Monroe <ian at monroe.nu> wrote:
> > Markey and I were talking about how we should perhaps drop Postgresql
> > and Sqlite, and just use MySQL embedded (in-process mysql support) and
> > perhaps support external MySQL as well.
> >
> > The primary advantage would be we'd have less to support. Just MySQL.
> > Supporting sqlite, mysql and postgresql has been a real burden. The
> > secondary advantage is the MySQL 5 is almost a real database and it
> > has a lot of features we could take advantage of. Views, probably more
> > consistency checking. Currently we are held back by the limited subset
> > of SQL that sqlite supports. Also recent versions of sqlite have been
> > having corruption fixes.
>
> I think postgres support is a really awesome thing to have, and I know
> of many people that use it. What is the advantage to dropping it,
> aside from the obvious cop out that it is less to maintain?
>
> Postgresql is more robust than Mysql, and requires virtually the same
> syntax (except for some more complicated statements). I'd be happy to
> keep postgres up to date.

Postgres sure is a good database (I only hear good things about it),
but maintaining multiple database backends puts us in the same
situation as maintaining multiple audio engines: We have to maintain
multiple code paths for very little return value. Risking stability.

I would wager that 70% of our users use the default DB backend
(SQLite), 25% use MySQL, and 5% use Postgres. With the upcoming
platform ports this percentage will shift even more to the default
backend. Cutting down on options will hurt a small fraction of our
users, but in the end it'll benefit the whole application.

-- 
Mark