[
https://issues.apache.org/jira/browse/LUCENE-3590?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13157924#comment-13157924
]
Robert Muir commented on LUCENE-3590:
-------------------------------------
BytesRef.copyBytes and equivalents are also heavily broken if offset != 0, and
the javadoc is non-intuitive:
{noformat}
* NOTE: this method resets the offset to 0 and resizes the reference array
* if needed.
{noformat}
This is ambiguous, its not obvious that it will reset offset to 0 *always* and
then copy, resizing if needed.
This is in comparison to e.g. append(), which is less broken because it at
least tries to append to the
existing offset, and if it needs to resize it resets offset to 0, but thats
fine as its a new byte[].
The former will overwrite unrelated data if bytesrefs are pointing to different
slices of the same array.
> minimize bytesref to be a ref to a byte[]
> -----------------------------------------
>
> Key: LUCENE-3590
> URL: https://issues.apache.org/jira/browse/LUCENE-3590
> Project: Lucene - Java
> Issue Type: Bug
> Affects Versions: 4.0
> Reporter: Robert Muir
> Priority: Blocker
> Attachments: LUCENE-3590_copyCtor.patch,
> LUCENE-3590_deprecate_comparators.patch, LUCENE-3590_equals_init.patch,
> LUCENE-3590_equals_init.patch, LUCENE-3590_with_clone.patch
>
>
> Setting this as blocker for 4.0, as this class is used in every API there.
> Currently this API is a little of a everything, sometimes its like a
> stringbuffer, it
> does numeric conversions, all kinds of stuff.
> We need this to be a ref to a byte[], nothing else. This other stuff can go
> somewhere else.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
