[
https://issues.apache.org/jira/browse/HBASE-1400?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12716072#action_12716072
]
Andrew Purtell edited comment on HBASE-1400 at 6/3/09 2:43 PM:
---------------------------------------------------------------
I notice when you tag something byte[] as XmlAttribute the result is base 64
encoding inside the attribute. I presume unmarshalling does the right thing.
How about:
{code:xml}
<CellSet>
<Row name="row name">
<Cell column="test:c1" timestamp="1128765413420141">
VGhpcyBpcyBhbiBleGFtcGxlIHZhbHVlLgo=
</Cell>
</Row>
</CellSet>
{code}
along with protobuf changes to match this structure?
One thing I can't figure out how to do is specify alternate encodings for cell
bodies. For example, before this was allowed:
{code:xml}
<Cell column="test:c1" timestamp="1128765413420141" encoding="base64">
VGhpcyBpcyBhbiBleGFtcGxlIHZhbHVlLgo=
</Cell>
<Cell column="test:c1" timestamp="1128765413420141" encoding="plain">
This is a plain text value!
</Cell>
{code}
which can represent significant savings if there is a large amount of plain
text data to transfer. It's arguable if this is really needed. I just mention
it.
was (Author: apurtell):
I notice when you tag something byte[] as XmlAttribute the result is base
64 encoding inside the attribute. I presume unmarshalling does the right thing.
How about:
{code:xml}
<CellSet>
<Row name="row name">
<Cell column="test:c1" timestamp="1128765413420141">
VGhpcyBpcyBhbiBleGFtcGxlIHZhbHVlLgo=
</Cell>
</Row>
</CellSet>
{code}
along with protobuf changes to match this structure?
> Improve REST interface semantics and efficiency
> -----------------------------------------------
>
> Key: HBASE-1400
> URL: https://issues.apache.org/jira/browse/HBASE-1400
> Project: Hadoop HBase
> Issue Type: Improvement
> Components: rest
> Reporter: Andrew Purtell
> Priority: Minor
> Attachments: stargate-0.0.1.zip, stargate-testing.pdf, stargate.pdf
>
>
> Improve the semantics of the REST interface: more metadata operations, bulk
> updates, protobufs (if Accept equals "application/x-protobuf" for GET or
> Content-Type equals the same for PUT or POST) instead of multipart/related
> (which is not supported now anyway) etc. for general efficiency and support
> for queries or scanners that return multiple KeyValues.
> I am working on a proposal.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
