lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hoss Man (JIRA)" <>
Subject [jira] Commented: (LUCENE-1727) Order of stored Fields not maintained
Date Thu, 02 Jul 2009 21:07:47 GMT


Hoss Man commented on LUCENE-1727:

bq. It's a good reminder to avoid documenting how something works as a guarantee that it will
always work that way.

I remember it being pretty heavily advertised as a feature for as long as i can remember --
it was the entire basis for adding the FieldSelector API.

Based on McCandless comments in email, it sounds like order was only ever maintained for fields
that don't use term vectors -- in which case the documentation was only ever partially correct.
 If it's possible to fix it to work at least as well as it use to that seems worth while considering
how much FieldSelector depends on it.

(I admit however: it's kind of scary that it's been such an implicit assumption but apparently
never had a unit test)

> Order of stored Fields not maintained
> -------------------------------------
>                 Key: LUCENE-1727
>                 URL:
>             Project: Lucene - Java
>          Issue Type: Bug
>          Components: Index
>    Affects Versions: 2.4, 2.4.1
>            Reporter: Hoss Man
>            Assignee: Michael McCandless
>             Fix For: 2.9
> As noted in these threads...
> somewhere prior to Lucene 2.4.1 a change was introduced that prevents the Stored fields
of a Document from being returned in same order that they were originally added in.  This
can cause serious performance problems for people attempting to use LoadFirstFieldSelector
or a custom FieldSelector with the LOAD_AND_BREAK, or the SIZE_AND_BREAK options (since the
fields don't come back in the order they expect)
> Speculation in the email threads is that the origin of this bug is code introduced by
LUCENE-1301 -- but the purpose of that issue was refactoring, so if it really is the cause
of the change this would seem to be a bug, and not a side affect of a conscious implementation
> Someone who understands indexing internals should investigate this.  At a minimum, if
it's decided that this is not actual a bug, then prior to resolving this bug the wiki docs
and some of the FIeldSelector javadocs should be updated to make it clear what order Fields
will be returned in.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message