Mailing-List: contact avro-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: avro-dev@hadoop.apache.org
Message-ID: <1768845276.1243633965811.JavaMail.jira@brutus>
Date: Fri, 29 May 2009 14:52:45 -0700 (PDT)
From: "Doug Cutting (JIRA)" <jira@apache.org>
To: avro-dev@hadoop.apache.org
Subject: [jira] Updated: (AVRO-2) optimize RPC handshake
In-Reply-To: <176308370.1239316513083.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/AVRO-2?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Doug Cutting updated AVRO-2:
----------------------------

    Attachment: AVRO-2.patch

This patch has ballooned a bit.

- The RPC handshake schemas are pre-processed by the m4 macro processor.   They're then compiled by Avro's schema compiler and the resulting Java classes are used in the RPC code.  This involved a lot of changes to build.xml.
 - To make this easier I renamed schema files to be .avsc and protocol files to .avpr, so that we can automatically use the right compiler.
 - To some degree the m4 stuff also fixes AVRO-19.
 - I fixed some bugs in the test code, where @BeforeMethod was used when @BeforeClass should have been used, so that RPC tests now reuse connections, as intended.
 - I fixed some bugs in the RPC server code where connections were not properly closed.
 - I fixed a bug in the specific compiler where code for a type could be generated multiply.
 - I added an RPC handshake test where the client uses a different server than the server, with an added method parameter.


> optimize RPC handshake
> ----------------------
>
>                 Key: AVRO-2
>                 URL: https://issues.apache.org/jira/browse/AVRO-2
>             Project: Avro
>          Issue Type: Improvement
>          Components: spec
>    Affects Versions: 1.0
>            Reporter: Doug Cutting
>            Assignee: Doug Cutting
>         Attachments: AVRO-2.patch
>
>
> In the current RPC handshake, each client sends its full protocol at the start of each session, and the server must return its full protocol.  This could be expensive, and should be avoided when possible.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.