lynx-dev Re: bug #25947, $http_proxy + /etc/lynx.cfg http_proxy

[Patrik Rak]

On Mon, 30 Aug 1999, Klaus Weide wrote:

> I have never seen something like this, certainly not if the *_proxy
> variables/options are syntactically correct.  Note that the trailing '/'
> is required.

I see. So this is the problem, obviously.

> Howwever, I don't really understand the description.  What is a "cache
> name"?  At what stage does http://www.debian.org/index.html "become"
> http://www.debian.org/www.debian.org/index.html, what makes you think it
> has become that?  How was the URL entered in the first place?  (For
> example, did you use full URLs including the "http://";?)

Here is the related excerpt from lynx.cfg:

http_proxy:http://proxy.ein.cz:3128/
https_proxy:http://proxy.ein.cz:3128/
ftp_proxy:http://proxy.ein.cz:3128/
gopher_proxy:http://proxy.ein.cz:3128/
no_proxy:ein.cz

everything else is as per default, and there is no user's .lynx.cfg.

If there is no $http_proxy, things work fine.

If I set it to http://proxy.ein.cz:3128, it does not work,
regardless if I specify the URL on commandline or via the G)o command.
For example, say I typed "lynx www.seznam.cz". Here is what squid sees:

936018384.508    944 192.168.3.129 TCP_MISS/404 346 GET 
http://www.seznam.cz/www.seznam.cz/ - DIRECT/www.seznam.cz text/html

However, if I set it to http://proxy.ein.cz:3128/, as you suggest above,
it works again. But other programs don't insist on having the
trailing slash in order to work, and lynx should follow, IMHO.

But thanks for the hint how to work around this anyway.

Patrik

P.S. Sorry about the previous "incomplete" bug report, I thought that
something like duplicating the hostname should be easy to spot in the
sources...