Tim,
Thank you.
>It can be done in SQL: "SUM(kogus) OVER (PARTITION BY toode
>ORDER BY ID) - kogus" (*1) will give you the running sum of
>the product up to that row. You can then subtract that value
>from the delivered quantity to calculate the delivered quan-
>tity for the current row.
> But doing so automatically is probably bad. For example,
>if a user has a purchase order with one position of two
>pieces and one position of four, it is very likely that when
>a shipment of four pieces arrives, the latter position shall
>be marked as delivered. So I would leave the decision to the
>user.
If four pieces arrived, first position of 2 pieces should marked as
delivered.
Second position of 4 pieces shoudl be marked as partialli delivered by
setting undelivered quantity
of this row to 2
How to use your suggestion for this ?
How to implement this is PostgreSql 8.1,8.2, 8.3 ?
Andrus.
> (*1) In PostgreSQL 9.0, you might be able to use "ROWS
BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING" instead
(untested).
PS. If possible please use cc: with my email address in reply.