Re: lynx-dev User Agent

[Klaus Weide]

On Mon, 23 Aug 1999, John Hawkinson wrote:

> Continuing somewhat of an old thread; today I noticed that
> www.movielink.com (aka www.moviefone.com) seems to give lynx a 0-line
> text/html page, whereas changing the user-agent fixes that.
> 
> Is there a normal protocol for resolving these issues with 
> web site maintainers? Since this is a high-visibility site,
> I thought I'd inquire if there was a history.

The only protocol I know of is BOB:  Bitch or Boycott.

> What would folks think of a feature that allowed one to associated
> a user-agent name with a particular web-site, so that one could
> more easily access recalcitrant web sites without annoying
> manual configuration?

I don't think much of it.  All this user-agent manipulation stuff is
trying to apply a technical spolution to a social problem.  It doesn't
really solve the problem.  It encourages trying to sneak around a block
that was put in place intentionally, until you get noticed.  Instead
of using the BOB protocol in the first place.

Not that this may not be woethwhile sometimes.  But does it belong in
lynx?  If you have enough time to paly around with user-agent strings
until you find one that "works", you should also have enough time to set
up junkbuster os another proxy like that.  The advantage is that all
programs can benefit, not just Lynx.

Having said all that, one place that would seem logical to hook this kind
of protocol-level stuff (URL-specific headers) into is the cernrules
mechanism, if you want to do it.  That already provides a framework for
doing things based on (limited) URL matching, you wouldn't have to
reinvent so much (keeping lists of URLs, option syntax and parsing etc.)

   Klaus