[
https://issues.apache.org/jira/browse/LUCENE-6577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15555730#comment-15555730
]
Michael McCandless commented on LUCENE-6577:
--------------------------------------------
bq. Does these checks be valid on z/OS (big-endian) ?
[~bneelima84]: yes ... Lucene processes the bytes for checksum purposes in its
own order, regardless of the byte order of the underlying platform.
But then, Java is "write once test everywhere" and I don't know of specific
tests on z/OS in the past ...
> Give earlier and better error message for invalid CRC
> -----------------------------------------------------
>
> Key: LUCENE-6577
> URL: https://issues.apache.org/jira/browse/LUCENE-6577
> Project: Lucene - Core
> Issue Type: Task
> Reporter: Robert Muir
> Fix For: 5.3, 6.0
>
> Attachments: LUCENE-6577.patch
>
>
> From digging into LUCENE-6576:
> We use a 32-bit CRC but store it in 64-bits, but today 32 bits are unused. We
> should check this on both read and write and if bits are invalid (because
> something is wrong in the JVM, or because on-disk stuff just got corrupted in
> that way), deliver a good exception.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
