W3C home > Mailing lists > Public > www-lib@w3.org > October to December 1998

Re: Reusing channel problem?

From: Henrik Frystyk Nielsen <frystyk@w3.org>
Date: Tue, 24 Nov 1998 17:39:29 -0500
Message-Id: <3.0.5.32.19981124173929.00ae59b0@localhost>
To: olga@goliath.eai.com, www-lib@w3.org
At 16:10 11/23/98 -0600, olga wrote:

>I have the problem with dropped connections(?):If downloading from some host
>fails then any other subsequent download from the same host also fails.
Loading
>from other servers are OK. This is not a cache problem since if I stop
>application and start it again then everything work (until first failure). 

Hmm, libwww goes through a lot of efforts in order to reuse the same TCP
connection (a channel is essentially an abstraction for a transport layer
like TCP). If possible, it tries to pipeline the requests in order to save
time.

Unfortunately some servers are written very poorly and don't play the game
of persistent connections well. Libwww has quite a lot of code for trying
to recover a dropped connection but I would have to look at the traces to
see what goes wrong - some things are impossible to recover from.

Do you have a test setup that I can try?

>When I looked at source code I saw that there is some notion of Channel
and that
>the channel is reused. Does it have to do with subsequent loadings from
the same
>server? If it does, could somebody shortly explain what it channel, how the
>channel could fail and how and where the checking that reusing the channel
is OK
>is done?

When the TCP connections hasn't been dropped in bad ways.

Henrik
--
Henrik Frystyk Nielsen,
World Wide Web Consortium
http://www.w3.org/People/Frystyk
Received on Tuesday, 24 November 1998 17:39:36 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 17:59:22 EDT