well,
1) it would be additional logic overhead to generate the unique id and keep
it global for all data providers
2) I doubt that ordering on 2 fields like "time" up to sec (or even to min)
and "integer" will be quicker when sorting using just one "long"
Scott Farquhar
<[EMAIL PROTECTED] To:
[email protected]
com> cc: Lucene Users List
<[email protected]>, (bcc: Iouli
Golovatyi/X/GP/Novartis)
06.04.2005 23:51 Subject: Re: Sorting date
stored in milliseconds time
Please respond to
java-user Category:
|-------------------------|
| ( ) Action
needed |
| ( ) Decision
needed |
| ( ) General
Information |
|-------------------------|
On Wed, Apr 06, 2005 at 01:02:35PM +0200, [EMAIL PROTECTED]
wrote:
> I'm forced to keep date up to milisec. The reason is simple: I get at
> least a couple of new messages per sec, if all of them are stamped with
the
> same time, the retrieval order id undefined, i.e. once I get it, let's
> say, as the last reference on the first page, other time - as the first
one
> on the second page. In case You do not keep the cache and always performs
> the same query even to get the second page it's getting even worse, i.e.
> displaying the messgaes 41-80, may bring the message 41, which You
already
> displayed on your first page as message 40.
Why can't you have a secondary sort on message id? Then you sort by time
and message id will always result in consistent ordering.
Cheers,
Scott
--
ATLASSIAN - http://www.atlassian.com
Expert J2EE Software, Services and Support
-------------------------------------------------------------
JIRA Docs: http://www.atlassian.com/software/jira/docs/latest
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
______________________________________________________________________
The Novartis email address format has changed to
[EMAIL PROTECTED] Please update your address book
accordingly.
______________________________________________________________________
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
