hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Matt Corgan (Commented) (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-5720) HFileDataBlockEncoderImpl uses wrong header size when reading HFiles with no checksums
Date Sat, 07 Apr 2012 23:36:17 GMT

    [ https://issues.apache.org/jira/browse/HBASE-5720?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13249415#comment-13249415
] 

Matt Corgan commented on HBASE-5720:
------------------------------------

The ideal way to split code in my opinion would be to put it in multiple source directories
in the same project, and then configure the build path to set up one-way visibility between
source directories.  For example, the src/test/java folder would be able to see the src/main/java
folder but not vice versa.  Unfortunately, eclipse does not support that.  It has a global
classpath per project.  Can't find the exact discussion, but similar to: https://bugs.eclipse.org/bugs/show_bug.cgi?id=224708.
 

I can think of two alternatives:

1) Put multiple src directories in the main project, so everything is in that project's classpath.
 Periodically have a human or something else remove non-core src directories from the classpath
and see if that causes compilation errors in the main src.  I just tried removing the src/main/test
directory from the classpath and was happy to see that the src/main/java folder still compiles.

2) A more heavy handed approach would be to create separate projects so you can actually enforce
unidirectional classpath visibility.  That's how I've been developing HBASE-4676 which can
be added to the main project as a separate jar.  Main project still compiles without it. 
The downside of this is that having multiple projects is more configuration overhead, like
you would probably need multiple svn repos.

I think moving towards #1 is a good start for core stuff while more exotic contributions could
be made by submitting jars.  I'll try to nudge us in this direction with HBASE-4676, just
wanted to make sure people agree that it's important =)
                
> HFileDataBlockEncoderImpl uses wrong header size when reading HFiles with no checksums
> --------------------------------------------------------------------------------------
>
>                 Key: HBASE-5720
>                 URL: https://issues.apache.org/jira/browse/HBASE-5720
>             Project: HBase
>          Issue Type: Bug
>          Components: io, regionserver
>    Affects Versions: 0.94.0
>            Reporter: Matt Corgan
>            Priority: Blocker
>             Fix For: 0.94.0
>
>         Attachments: 5720-trunk-v2.txt, 5720-trunk.txt, 5720v4.txt, 5720v4.txt, 5720v4.txt,
HBASE-5720-v1.patch, HBASE-5720-v2.patch, HBASE-5720-v3.patch
>
>
> When reading a .92 HFile without checksums, encoding it, and storing in the block cache,
the HFileDataBlockEncoderImpl always allocates a dummy header appropriate for checksums even
though there are none.  This corrupts the byte[].
> Attaching a patch that allocates a DUMMY_HEADER_NO_CHECKSUM in that case which I think
is the desired behavior.

--
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

        

Mime
View raw message