Re: lynx-dev dev.24 Archive link runaround

[Greg Marr]

At 09:30 PM 4/26/99 , Philip Webb wrote:
>990426 Greg Marr wrote: 
>> Say in  http://www.flora.org/lynx-dev/index.html
>> there is a link to "about-lynx.html".
>> If you entered the URL  http://www.flora.org/lynx-dev/
>> and then selected that link,
>> Lynx would ask for  http://www.flora.org/lynx-dev/about-lynx.html .
>> Now say you entered the URL  http://www.flora.org/lynx-dev
>> and the server didn't find a file called  lynx-dev ,
>> but did find a directory called  lynx-dev/
>> and so sent back the contents of  index.html  in that directory.
>> As far as Lynx knows, you are looking at
http://www.flora.org/lynx-dev ,
>> so when you select that link,
>> it will ask for  http://www.flora.org/about-lynx.html
>> instead of  http://www.flora.org/lynx-dev/about-lynx.html .
>
>thanx for this very clear explanation,
>which deals with a real-life example, not dogmas or doctrines.
>this problem about relative URLs hasn't been mentioned before,
>but if the client makes your assumption, things will get confused.

The client has no choice but to make the assumption.  It asked for a
document, and got it.  It has no possible way to know that the URL of
what it received was anything other than what it asked for.

>> What the server can do, when you've entered an *invalid* URL,
>> is return a `Moved Permanently' status message redirecting you
>> to  http://www.flora.org/lynx-dev/ .  Lynx will then disconnect
>> from the server, reconnect to the same server, and ask for that URL
>> Now when it gets the page, it knows how to properly handle relative
URLs.
>
>so Flora could send that message:
>one wonders why it has been set up to do what it does instead.

It may be as simple as an error in a configuration file.

>>  http://www.perl.com/CPAN  is a file
>> that gives you a list of all the CPAN servers and has a CGI backend
>> that allows you to select the one you want to use when accessing the
CPAN.
>>  http://www.perl.com/CPAN/  and
>>  http://www.perl.com/CPAN/blah/blah/blah.html
>> redirect you to a server within the CPAN network
>> that seems to be the best choice for you at the time.
>> In the case of the second URL, it also brings you to the location
>> of  blah/blah/blah.html  within that server's CPAN archive.
> 
>yes, i understood that,
>but the fact that the added  /  causes redirection seems arbitrary:
>how are users supposed to know how these URLs will differ in their 
>effects?

They are told that they do:
"Welcome to the CPAN multiplexer. Whenever you ask for a path from this
site that begins http://www.perl.com/CPAN/ you will automatically get a
site close to you. Since you're reading this now, you must have called
it without a slash, which gives you the chance to pick which site you
prefer."

There is other documentation about this, but I can't find it right now,
and it's not really relevant to this discussion.

-- 
Greg Marr
address@hidden
"We thought you were dead." 
"I was, but I'm better now." - Sheridan, "The Summoning"