Re: [Gnumed-devel] gmBusinessDBObject._xmins_refetch_col_pos

[Karsten Hilbert]

On Thu, Dec 23, 2004 at 01:15:00AM +1100, Ian Haywood wrote:
> 
> Karsten Hilbert wrote:
> 
> >>self._idx.index ()
> >
> >Uhm, OK, _refetch_payload pulls, say, 6 fields, the last of
> >which (eg idx=5) being xmin_something. The index to that query
> >becomes self._idx.
> No, we use the *original* self._idx which matches our original
> self._payload
What I was describing IS the original self._idx.

> so self._idx.index () transforms the field names
Yes, for the query in _cmd_refetch_paylod.

> (which we insist are the same across the two queries)
Ah, there's the rub. That would mean you'd need to pull dummy
fields in the last query of _cmds_store_payload to match up
actual positions to positions in self._idx.

> >_store_payload in the very last query pulls xmin_something
> >again - but now it would have the index 0 in the result. 
> Indeed, we ignore those positions.
And how do you then know which position is which field ?
Certainly not by the original self._idx unless one pulls dummy
columns.

Karsten
-- 
GPG key ID E4071346 @ wwwkeys.pgp.net
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346