[Babase] New Ranker login and misused groomings

Karl O. Pinc kop at meme.com
Wed Apr 9 17:49:37 EDT 2008


On 04/09/2008 04:29:58 PM, Jun Yang wrote:
> I see.  That should be a very simple fix. Currently, the query that
> retrieves the interactions doesn't filter out any interactions at all.
> We just need to put in the appropriate filters. The questions is how
> flexible we should make it though:
> 
> * We can hard-code the filter to count agonistic interactions only---I
> assume the condition would be CLASS = 'A'? Or should it be ACT='AS' OR
> ACT='DS' OR ACT='OS'?
> 
> * We can let user pick which subset of ACTs to include in counting
> (e.g., AS+DS, AS alone, AS+DS+OS, etc.). This makes the whole thing a
> bit messier but certainly would be possible.
> 
> What do you guys think?

The tradeoff the old ranker program made was to
select on a single CLASS = ''.  I'd say if you want
to get more complicated than that just give up and
force the user to enter his own SQL statement that
would select the desired ACTS rows.  Stick that
query into a subquery of your actual SELECT that selects
the interactions: WHERE acts.act IN (select-acts.act-from-
whatever-else-the-user-wrote-for-a-select)

Dunno...


Karl <kop at meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein



More information about the Babase mailing list