[Bioperl-l] DB2 driver for BioPerl

[Jonathan Crabtree]

Hi Florian,

On Thu, Jul 2, 2009 at 10:52 AM, Florian Mittag
<florian.mittag at uni-tuebingen.de> wrote:
> Well, what made me suspicious was that the returned columns were exactly the
> ones from the term table plus the NULL column. One way to verify this would
> be to look whether the same thing happens with other tables as well.

Others may disagree, but I think it's fairly clear (from just looking
at the subroutine you mentioned) that the inclusion of the NULL value
is most definitely deliberate; note that it is only done if $entitymap
doesn't have a value for the table in question, and $entitymap is
described as follows:

A reference to a hash table mapping entity names to aliases (if
omitted, aliases will not be used, and SELECT columns can only be
from one table)

So I suspect what we're seeing here is a select in which aliases
_aren't_ being used and therefore the order of the returned values is
significant, and the NULL value is needed to keep everything in the
right order for whatever piece of code is reading the result.  But
never having worked much with BioSQL I don't know where you'd go to
find the type information needed to determine what type the NULL value
needs to be coerced into...

Jonathan