[jira] [Commented] (LUCENE-6090) don't wrap Deflater in CompressionMode

["Adrien Grand (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LUCENE-6090?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14232782#comment-14232782 ] 

Adrien Grand commented on LUCENE-6090:
--------------------------------------

I am wondering if this dummy byte is actually required. Although the javadocs state so, I found this comment from Mark Adler who says that it was only required until zlib 1.1.4: http://stackoverflow.com/questions/9770364/decompressing-a-string-using-java-util-zip-inflater (see 2nd answer).

If we want to keep it for conformity with javadocs, can you put a comment that Inflater requires it?

> don't wrap Deflater in CompressionMode
> --------------------------------------
>
>                 Key: LUCENE-6090
>                 URL: https://issues.apache.org/jira/browse/LUCENE-6090
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: Robert Muir
>         Attachments: LUCENE-6090.patch, LUCENE-6090.patch
>
>
> This outputs zlib header/footer and computes adler32 for each block. The space is nothing, but the adler32 computation on encode/decode has a cost, and we already have our own checksum.
> Since we currently compress/decompress at merge, this reduces the overall time of merging stored fields with deflate vs lz4, from 1.8x to 1.5x, reducing some of the pain.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org