Hi,
I have been looking into Fastbit's support of aggregate functions and I have a
few questions on that (or probably they turn out to be all feature requests):
1. is there currently any way to count the number of distinct values in a group?
In some SQL products, I'd issue something like
SELECT event_id, /* id of the event */
COUNT(*) /* how many times did the event occur */,
COUNT(DISTINCT(item_id)) /* how many distinct items where affected by
the event */
FROM table
GROUP BY event_id;
I have peeked at the ibis source code and did not find any equivalent for a
COUNT(DISTINCT()) or a DISTINCT.
2. Native Fastbit support for more aggregate functions like STD(), VAR() would
be great.
3. Conditional logic for aggregate functions would be absolutely great to
create cross-breaks.
In some SQL products, this would look like
SELECT event_id, SUM(IF(event_type = 1, 1, 0)), SUM(IF(event_type = 2, 1, 0))
FROM table
GROUP BY event_id
4. Finally, HAVING clauses operating on the results of grouped values would be
a nice-to-have extension to Fastbit though not absolutely necessary (can be
implemented outside of Fastbit).
For any of the above features, do you know if there is an existing Fastbit
workaround? If not, is anything of the above on the Fastbit development agenda?
Thanks in advance for letting me know and best regards
Jan
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
