qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Aidan Skinner (JIRA)" <qpid-...@incubator.apache.org>
Subject [jira] Commented: (QPID-1251) IO transport does not cleanly close on Windows
Date Mon, 18 Aug 2008 15:13:44 GMT

    [ https://issues.apache.org/jira/browse/QPID-1251?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12623369#action_12623369

Aidan Skinner commented on QPID-1251:

  [4:07 pm] partychat0  ["rhs"] I think the safest thing to do is have IoSender replace the
socket.shutdownInput() with a socket.close() on windows only. 
 [4:07 pm] partychat0  ["rhs"] IoReceiver rather. 
 [4:07 pm] partychat0  ["rhs"] This will cause spurious exceptions to be reported. 
 [4:07 pm] partychat0  ["rhs"] But I'd rather report spurious exceptions than swallow non
spurious ones. 
 [4:08 pm] aidan.x.skinner  basically close() would always throw something?
  [4:09 pm] partychat0  ["rhs"] I think the exception would probably be reported to the connection's
exception listener actually. 
 [4:10 pm] partychat0  ["rhs"] Assuming one is set. 
 [4:10 pm] aidan.x.skinner  that's not too bad
 [4:11 pm] partychat0  ["rhs"] But it would probably be best to just try it and see what happens.

> IO transport does not cleanly close on Windows
> ----------------------------------------------
>                 Key: QPID-1251
>                 URL: https://issues.apache.org/jira/browse/QPID-1251
>             Project: Qpid
>          Issue Type: Bug
>          Components: Java Client
>    Affects Versions: M3
>         Environment: Windows
>            Reporter: Arnaud Simon
>             Fix For: M3
> Description: 
> There is an issue on Windows platform with the io transport as socket.shutdownInput()
does not result of the reader getting an EOF. This means that the IOReceiver thread does not
terminate as expected and the main thread does not join. This prevents the protocol version
switching mechanism to work.  
> Solution:
> We will be using a new nio transport that should solve this issue. I would therefore
suggest that we only provide a simple mechanism for selecting the default version used and
tell Windows users to set it to 0.9 when using the current java broker. 

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

View raw message