[Asterisk-Dev] Current database abstraction effort ?

Florian Overkamp florian at obsimref.com
Mon Jan 5 05:37:35 MST 2004


> -----Original Message-----
> > However, I'm wondering whether adding unixodbc requirement to do
> > database handling may :
> > - make building asterisk harder (more packages to get)

Actually, with current state of things, Asterisk will just detect it. If you
don't have odbc, it will build without. Ofcourse, it would make sense to
install odbc if you wish to use it :-)
So yes there are more packages to get _if you want that functionality_.

> > - make it more prone to various lib incompatibilites (is unixodbc
> >   compatible between versions ?)

No clue.

> > - add some unnecessary complexity, and/or have significant a
> >   performance cost. 

Complexity: Yes, a little at setup. But as you said it can be learned. With
mysql being separated into asterisk_addons I would say that mysql support is
more complex than odbc support.

Performance: I am sure there is a cost compared to direct database access.
However, it performs well enough for my purpose, so I don't see an issue. I
believe BKW has done some performance testing in the past ?

> > - not provide that much of an help : how well is unixodbc (and its
> >   database drivers) packaged in the most widespread distros
> >   (Debian, RH/Fedora, Suse, MDK...) ? 

My debian machines (stable) have all required odbc stuff in nice packages.
Works like a charm.

> > Could the people familiar with unixODBC share their views on these
> > topics ?

I'm not really familiar with odbc, My experiences are based on the two
machines I migrated to odbc logging just last week. However, I must say I
had a harder time getting postgresql to work than unixodbc with mysql :-)


More information about the asterisk-dev mailing list