hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ortwin Glück <...@odi.ch>
Subject Re: [Fwd: Re: [HttpClient] Patched archive]
Date Fri, 08 Jul 2005 09:20:29 GMT

> Hi Ortwin,
> 
> I will let you know tomorrow if the patch you sent works in my setup.
> I have anyways an offtopic question regarding your statement taken from
> bugzilla
> <quote>
> Yes HttpClient uses java.net. No, HttpClient does not use SocketChannel
> (nio)
> nor Socket.connect with a timeout since those APIs appeared in JDK 1.4.
> HttpClient is however compatible with JDK 1.3.1.
> </quote>
> 
> In the light of this statement, how does the new abort feature handle
> limit situations of TCP state? AFAIK it is not possible to shut down
> imediately and in a clean way a socket before NIO. I have seen
> situations where a thread blocked in socket write operation (partial
> network failures, tcp buffer full) was imposible to be interrupted by
> other thread in any way (by calling socket close or other ways).
> So, what would happen if one thread executes a http method, and
> another tries to cancel it with an abort call? (hit escape in firefox
> kind of scenario).
> 
> Regards and thank you for your time,
> Horia

Horia,

HttpMethod.abort just calls HttpConnection.close. This in turn first
closes output and input streams and the closes the Socket. In my
experience this has never caused any problems. We have never received
notice from anybody that an abort was not immediate.

Ortwin Glück

---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: httpclient-dev-help@jakarta.apache.org


Mime
View raw message