W3C home > Mailing lists > Public > www-lib@w3.org > January to March 1999

Re: Fw: libwww flow control?

From: Raffaele Sena <raff@nuvomedia.com>
Date: Tue, 2 Mar 1999 14:19:33 -0800
Message-ID: <00bc01be64fa$bfef86a0$52c0a8c0@tux.nuvomedia.com>
To: "Peter Chan" <pchan@netgravity.com>, "www-lib" <www-lib@w3.org>, "Henrik Frystyk Nielsen" <frystyk@w3.org>
>
>Hmm, I start to see something similar - I have some 902 errors as well.
>Could I get you to take a careful dump with lots of traces (especially
>stream, protocol, and core) and see whether you can find something (you can
>also send them to the list, of course).
>
>Henrik
>


I used to have these 902 all the time, now after applying your latest
changes
they changed into status = -1 in the terminate handler.

I did some digging and find out that at one point a request was leaving the
host
locked and all the successive requests to that host were failing with -1.

I applied the following patch:

cvs diff HTHost.c (in directory c:\SOFTWAREDEV\LIBWWW\Library\src\)
Index: HTHost.c
===================================================================
RCS file: /sources/public/libwww/Library/src/HTHost.c,v
retrieving revision 2.61
diff -r2.61 HTHost.c
1109a1110
>  host->lock = HTList_firstObject(host->pending);

Basically, after updating the pending list (by removing a net object), the
lock
is updated with the (possibly) new net object (maybe just setting host->lock
to
NULL if host->lock == net would have been enough).

Now, after the point where my application was failing, I can get every other
request.

I would think that, if my patch is correct, it should be applied to all the
code
that updates the host->pending list (mostly for deleting entries), but I'm
not so
familiar with the code to be sure that's really the right think to do.

-- Raffaele
Received on Tuesday, 2 March 1999 17:19:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 23 April 2007 18:18:29 GMT