[
https://issues.apache.org/jira/browse/HBASE-6226?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13404618#comment-13404618
]
Matt Corgan commented on HBASE-6226:
------------------------------------
Thanks guys. I left a reply on the encoder instantiation via reflection
discussion. repeating here:
{quote}
These encoder implementations are singletons that don't even have any member
fields, so they are as lightweight as can be. I think it's better they
fail-fast, though all implementations are currently bundled with hbase so it'd
be weird that they ever fail.
The only reason i made any change is that they are no longer in the
compilation-time classpath making the existing instantiation impossible. So one
concern might be that the implementation could change name or package, but i'm
thinking unit tests cover us well there.
The weird thing now is that hbase-common knows about different implementations,
but it mirrors how the Compression implementations are handled. Maybe in the
future we can consider having the HColumnDescriptor take these fully-qualified
class names so that different implementations can be plugged without being
added to this enum, but that's veering off track for this jira.
{quote}
Stack - it looks like Phabricator correctly matches up moved files, but if
nothing in the file has changed it makes it look like a new file. Some of your
comments on that top file are are things that already exist on trunk. All good
points, but maybe we can address separately.
> move DataBlockEncoding and related classes to hbase-common module
> -----------------------------------------------------------------
>
> Key: HBASE-6226
> URL: https://issues.apache.org/jira/browse/HBASE-6226
> Project: HBase
> Issue Type: Improvement
> Components: io, regionserver
> Affects Versions: 0.96.0
> Reporter: Matt Corgan
> Assignee: Matt Corgan
> Fix For: 0.96.0
>
> Attachments: 6226-suggestion.txt, HBASE-6226-v1.patch,
> HBASE-6226-v2.patch, HBASE-6226-v3.patch, HBASE-6226-v4.patch
>
>
> In order to isolate the implementation details of HBASE-4676 (PrefixTrie
> encoding) and other DataBlockEncoders by putting them in modules, this pulls
> up the DataBlockEncoding related interfaces into hbase-common.
> No tests are moved in this patch. The only notable change was trimming a few
> dependencies on HFileBlock which adds dependencies to much of the
> regionserver.
> The test suite passes locally for me.
> I tried to keep it as simple as possible... let me know if there are any
> concerns.
--
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
