[
https://issues.apache.org/jira/browse/HCATALOG-332?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13245573#comment-13245573
]
Francis Liu commented on HCATALOG-332:
--------------------------------------
{quote}
So this is a semantics question. If we say HCatInputFormat is public but
HCatBaseInputFormat is private, I guess that makes sense. Both must be stable
though, because HCatBaseInputFormat defines part of HCIF's interface.
{quote}
That's fine, I only want to prevent users from subclassing the base class since
this is something internal.
{quote}
InputJobInfo.getPartitions() returns a list of PartInfos. InputJobInfo is
stable. It seems meaningless to declare InputJobInfo to have a stable interface
while threatening to evolve classes used in that interface. My assertion is
that stability is cumulative. No part of a stable interface can be less than
stable.
{quote}
If we decide to follow that semantic then anything using PartInfo shouldn't be
stable. As it will be changing soon. HCATALOG-308 alluded to the fact that we
are missing a partition descriptor class and some of that class' responsibility
is currently being done by PartInfo.
> HCatalog needs interface classification
> ---------------------------------------
>
> Key: HCATALOG-332
> URL: https://issues.apache.org/jira/browse/HCATALOG-332
> Project: HCatalog
> Issue Type: New Feature
> Reporter: Ashutosh Chauhan
> Assignee: Alan Gates
> Fix For: 0.5
>
> Attachments: HCAT-332-2.patch, HCAT-332.patch
>
>
> All the public facing interfaces should be marked for intended audience and
> stability. Ref: HADOOP-5073
--
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
