[
https://issues.apache.org/jira/browse/LUCENE-9047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17248808#comment-17248808
]
Ignacio Vera commented on LUCENE-9047:
--------------------------------------
I am wondering if the best approach for this issue is to first create new
Lucene90 codecs for all the current codecs. The current codecs should be moved
to backwards codec together with a copy of org.apache.lucene.util.packed. So
what is important here is that the backward codecs should depend on a copy of
PackedInts that works on Big Endian.
Once this is done, it should be trivial to change the DataInput /DataOutput and
org.apache.lucene.util.packed to work on little endian. For backwards codec it
should be then possible just to wrap the created IndexInputs. The only place
that we might need to be careful is when opening segment files in
SegmentInfos#readCommit?
> Directory APIs should be little endian
> --------------------------------------
>
> Key: LUCENE-9047
> URL: https://issues.apache.org/jira/browse/LUCENE-9047
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Priority: Major
> Time Spent: 7h 10m
> Remaining Estimate: 0h
>
> We started discussing this on LUCENE-9027. It's a shame that we need to keep
> reversing the order of bytes all the time because our APIs are big endian
> while the vast majority of architectures are little endian.
--
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
