[BioSQL-l] UK definition for seqfeature

[Aaron J Mackey]

Of course some will argue that type_term and source_term should really
just be qualifier values, in a data-format agnostic viewpoint (and I've
never really understood the whole type/source dichotomy; my GFF files
are almost always wrong).

But I have no objection to making it NOT NULL DEFAULT "" (term as well).

-Aaron

On Mon, 17 Mar 2003, Hilmar Lapp wrote:

> There is currently a compound unique key defined on seqfeature
> comprising of (bioentry_id,type_term_id,source_term_id,rank).
>
> A while ago all but the first were nullable columns with the previously
> known problems of different behaviour between RDBMSs. The situation is
> much clearer defined now, as only source_term_id is left as nullable.
>
> Can anyone name reasons why we should leave this column nullable? I.e.,
> can people think of sensible use cases where you would not have a
> source_tag for a feature?
>
> I believe even in GFF source is mandatory ...
>
> 	-hilmar
>
>

-- 
 Aaron J Mackey
 Pearson Laboratory
 University of Virginia
 (434) 924-2821
 amackey at virginia.edu