[
https://issues.apache.org/jira/browse/LUCENE-9794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17289687#comment-17289687
]
Robert Muir commented on LUCENE-9794:
-------------------------------------
Looks great.
Yeah this compressing reader/writer is at an awkward place. Even worse is the
methods such as readZDouble/writeZDouble/readTLong/writeTLong which should be
easy to effectively test with unit tests directly but to my knowledge are not.
java and abstractions, too many abstractions...
Most current testing for any skipping logic here is probably via concrete test
subclasses extending
https://github.com/apache/lucene-solr/blob/master/lucene/test-framework/src/java/org/apache/lucene/index/BaseStoredFieldsFormatTestCase.java
. I am guessing that is what testReadSkip() is trying to check in a very
indirect way.
> Optimize skipBytes implementation in remaining DataInput subclasses
> -------------------------------------------------------------------
>
> Key: LUCENE-9794
> URL: https://issues.apache.org/jira/browse/LUCENE-9794
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Greg Miller
> Priority: Minor
> Attachments: LUCENE-9794.patch
>
>
> LUCENE-9480 introduced a more efficient method for byte skipping in
> IndexInput and its subclasses, but the rest of the DataInput implementations
> are still delegating to DataInput#skipBytesSlowly.
>
> This issue tracks optimizing the remaining DataInput skipBytes
> implementations. Here's the list remaining:
> * CompressingStoredFieldsReader (contains an anonymous DataInstance impl
> internally)
> * ByteSliceReader
> * ByteBuffersDataInput
> * InputStreamDataInput
> * PagedBytesDataInput
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
