[OpenSER-Devel] mismatch between sc.dbtext and textdb.sh

[Henning Westerholt]

On Monday 16 July 2007, Daniel-Constantin Mierla wrote:
> > Ovidiu Sas and i noticed during a discussion that there is a
> > inconsistency between the old dbtext infrastructure scripts "sc.dbtext"
> > and "textdb.sh". sc.dbtext creates the tables without a "id(int,auto)
> > field, textdb.sh (and the new infrastructure in svn trunk) add this
> > field.
> >
> > The new openserctl script (patch from Ovidiu, recently merged) don't work
> > at the moment because of this mismatch.
> >
> > So my question is:
> > What is the right behaviour? Are these fields necessary? Are there
> > evaluated in the dbtext module?
>
> the field is not in used by most of the modules. The auto-increment
> column was added mainly to help additional tools and have the primary
> key on an integer (faster db operations). If I am not wrong, msilo's
> table has been updated to use this column instead of old one.
>
> However, the id should be created to maintain consistency between table
> definitions across DB engines.

Hello Daniel,

Ovidiu mentioned some concerns regarding the additional memory usage, if this 
db entries are not used. But in my opionion consistency is more important and 
if also some modules starts to use the id column then this is another reason 
for the new schema.. 

I'll change the dbtext code in openserctl to the schema with id fields.

Cheers,

Henning