[
https://issues.apache.org/jira/browse/THRIFT-388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12865888#action_12865888
]
Jeff DeCew commented on THRIFT-388:
-----------------------------------
First, I think this proposed change is great.
I also think it would be great if the distinction between normal and oneway
calls were accessible at the TTransport level.
I've implemented a JMS based transport (in Java) which uses a heartbeat pattern
to allow clients to timeout appropriately if the server is not processing the
message, but because the transport has no access to the oneway flag, the server
sends an unnecessary heartbeat for oneway calls. This particular example is
not a big deal, but it does seem to make sense that the transport classes
(client AND server) should know ahead of time whether there will be a response.
> Use a separate wire format for async calls
> ------------------------------------------
>
> Key: THRIFT-388
> URL: https://issues.apache.org/jira/browse/THRIFT-388
> Project: Thrift
> Issue Type: Improvement
> Reporter: David Reiss
> Assignee: David Reiss
> Priority: Minor
> Fix For: 0.1
>
> Attachments:
> 0001-THRIFT-388.-Create-a-ONEWAY-message-type-that-is-a.patch
>
>
> If async calls look different from normal calls on the wire, then a proxy
> with no access to the particular IDL will be able to distinguish them. That
> will be good.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
