[
https://issues.apache.org/jira/browse/THRIFT-492?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wil Tan updated THRIFT-492:
---------------------------
Attachment: thrift_client-v2.diff
I'm using framed transport, and found that making the gen_server:call() timeout
infinity for reads fixes the issue for me, i.e. the timeout is determined by
the initial thrift_client:call(), so we can allow the underlying transport to
take as long as it wants.
If you think that this is patch is the way to go, then maybe other transports
would need similar fixes as well, but I don't use them so I'm not so
comfortable patching them.
> Add Timeout Argument to thrift_client:call()
> --------------------------------------------
>
> Key: THRIFT-492
> URL: https://issues.apache.org/jira/browse/THRIFT-492
> Project: Thrift
> Issue Type: Improvement
> Components: Library (Erlang)
> Environment: N/A
> Reporter: Wil Tan
> Priority: Minor
> Attachments: thrift_client-v2.diff, thrift_client.diff
>
>
> thrift_client:call() does not use a timeout option when calling
> gen_server:call(), so the default timeout of 5 seconds is used. It would be
> good to offer that to the user of the module.
> The attached trivial patch does exports a new function call/4 which does
> this, and makes the original call/3 function pass in the default 5000 ms
> timeout value.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
