Could you file a JIRA on this so we can track it and fix it?
Alan.
On Jan 5, 2010, at 7:29 AM, Vincent Barat wrote:
Without the limit, there is also the issue.
If only one of the 2 ORDER BY is performed, there is no issue.
Le 05/01/10 16:19, Mridul Muralidharan a écrit :
I had noticed that order by followed by limit (to do top K) has
issues.
Maybe same thing is applicable even to inner plans ?
Does it work with the limit removed ? (Just to test it, though
logically
it might be wrong).
Regards,
Mridul
Vincent Barat wrote:
Hello,
It seems to have a bug in PIG when ORDER BY is used with the DESC
modifier:
I have the following script:
imei_start = FOREACH sessions GENERATE imei, start;
imei_starts = GROUP imei_start BY imei;
imei_retained_period = FOREACH imei_starts {
ordered_imei_start = ORDER imei_start BY start DESC;
first_start = LIMIT ordered_imei_start 1;
rev_ordered_imei_start = ORDER imei_start BY start ASC;
last_start = LIMIT rev_ordered_imei_start 1;
GENERATE group, ordered_imei_start, rev_ordered_imei_start;
};
ordered_imei_start and rev_ordered_imei_start are actually the same
(they are both sorted in the ASC way).
Is it a known bug ?
