W3C home > Mailing lists > Public > www-jigsaw@w3.org > July to August 1997

Re: List of Jigsaw proxy problems

From: Yael Stav <stav@radview.com>
Date: Thu, 24 Jul 1997 16:33:36 +0300
Message-ID: <33D759B0.65840233@radview.com>
To: jigsaw mailing list <www-jigsaw@w3.org>
Following my previous message (enclosed below) about several problems
using Jigsaw
as a proxy, I would like to give some more information:

About problem no.2 :
================
(  'An HTTP error occured while getting
 http://www.entrepreneurmag.com/cgi-bin/nph-redirect.pl?SPONSOR_URL=http://www.entrepreneurmag.com&SPONSOR_ID=HOME.

    Details "expecting :(58) got (10) context:
Netscape-Communications/1.1 ". ' )

The HTTP response which is sent (probably by a non-perfect CGI) lacks
the "server:" field
label. The header received is something like:
"

Jigsaw can't overcome this malformed header and throws the exception
without
delivering the response (both Netscape and Explorer browsers overcome
it).

About problem no.3:
===============
(Null content-type in http://merchant-int.netscape.com)
In the HTTP response sent from the server the "content-type:" field
label is
replaced with "content:" label. This must be the reason for Jigsaw not
finding the
content-type of the document.

 Needless to say, common browsers do succeed in finding out the
content-type.

Can anyone help with a solution to these?
Is there any clue for the big problem stated in problem no.1?

Thanx.

Yael Stav wrote:

> Hi all,
> Using Jigsaw proxy (with no cache) I encountered several problematic
> URL's:
>
> 1.
>     a. http://customnews.cnn.com
>                  Netscape browser gets "Document contains no data"
>
>     b. http://www-il.radview.com
>                 Netscape browser gets "Document contains no data"
>
>     c. http://customnews.cnn.com/cnews/pna.myhome
>                 Netscape browser gets "Document contains no data"
>
>     In all cases getting these url's directly from Netscape gives
>     correct HTMLs.
>
>     A bit of digging exposed the fact that in all cases there was a
>     200 OK status accepted and probably complete headers but also an
>     IOException of "socket was closed".
>    Could this has to do something with zero content-length?
>
> 2.
>     http://www1.playboy.com/ads/entrepreneur2.cgi  (a bit embarracing)
>
>             Results (after a redirection) in:
>     'An HTTP error occured while getting
>
> 'http://www.entrepreneurmag.com/cgi-bin/nph-redirect.pl?SPONSOR_URL=http://www.entrepreneurmag.com&SPONSOR_ID=HOME.
>
>     Details "expecting :(58) got (10) context:
> Netscape-Communications/1.1 ". '
>
>     Without jigsaw proxy I get the first redirection, then another one
>
> to        http://www.entrepreneurmag.com" which gave me a correct
> HTML.
>     This problem is definately a MimeParser problem. After some more
> digging I found
>     out that it's source is the "expect" method from the
> "parse822HeaderName" method.
>
> 3.
>     http://merchant-int.netscape.com
>
>     Is received OK with or without Jigsaw but I discovered that
> instead
> of getting
>     content-type of 'text/html' Jigsaw gets null content-type. I guess
>
> it has something to do
>     with wrong parsing.
>
> I read a previous mail about the "Document contains no data" problem,
> which stated that it's a known bug in 1alpha5 Jigsaw but I couldn't
> find
> any further documentation about this bug, nor any
> way to solve it.
> I know these cases may be cases of non perfect HTTP, but I would like
> my
> proxy to be
> able to recover from such cases at least the way the Netscape browser
> does.
>
> I'm using JDK 1.0.2 and 1.1.2 on NT4.0 .
> Thank you for any help.
>  --
> Yael Stav
> RadView Software Ltd.      email : stav@radview.com
> Tel-Aviv, ISRAEL         voice:   +972-3-6459013
> http://www.radview.com     fax:     +972-3-6471406
Received on Thursday, 24 July 1997 09:34:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 9 April 2012 12:13:26 GMT