Re: Suggestion: ALTHREF attribute

James Green (jmkgre@essex.ac.uk)
Thu, 15 Jan 1998 13:43:25 +0000 (GMT)


From: James Green <jmkgre@essex.ac.uk>
To: Jordan Reiter <jreiter@mail.slc.edu>
Cc: Jukka Korpela <jkorpela@cc.hut.fi>, www-html@w3.org
In-Reply-To: <l0302091fb0e3a3d00fd2@[204.71.8.33]>
Message-Id: <SIMEON.9801151325.A@sf125.essex.ac.uk>
Date: Thu, 15 Jan 1998 13:43:25 +0000 (GMT)
Subject: Re: Suggestion: ALTHREF attribute


On Thu, 15 Jan 1998 06:31:46 -0500 Jordan Reiter <jreiter@mail.slc.edu> 
wrote:


> For the sake of simplicity and widest accessibility, I suggest listing all
> of the possible document links in the text, instead. In addition, there's
> something offering a solution of sorts at http://www.purl.org, I *think*.

Hm, seems slightly unclear as to it's purpose. I think what it's trying 
to do is have URLs point to it's directories, and for the PERL server 
to return the actual URL (a clear redirection). This does seem slightly 
pointless to me, as anything like that ought to be client-side, rather 
than putting strain on the server (you can image one server giving out 
all URLs for sites moved from one of those freebie web space servers!).

Whilst it does not provide a solution to a page full of external URLs, 
a better system to the one originally starting this thread would be a 
header tag telling the browser where a mirror of the site is located. 
For example, a domain-name could be as follows:

1st site at http://www.myname.com/

2nd site at http://www.myname.com.au/

The index.html of this first site would have a a header tag of the form:

<MIRROR HREF="http://www.myname.com.au/">

The 2nd site the same, but with the the 1st URL.


This invariably brings up the problem of what to do should the user 
bring up http://www.myname.com/content/articles/170.html which has an 
internal link to, say, 171.html, which is down or too slow. Should the 
MIRROR tag have the name URL as per the first example, and have the 
browser take a look and find the file within that domain directory, or 
should the MIRROR URL be of the form:

<MIRROR HREF="http://www.myname.com.au/conent/articles/">

in which case all the browser has to do is find the file. 

But what if the two directories are completely different up to your 
'area'?

For example (real one):

1) http://users.aol.com/cnsweb1/

mirrored at:

2) http://schools.sys.uea.ac.uk/schoolnet/jg/

Under such situation the URL mirror should point to the full path up to 
the root directory you own, e.g. the 1st site would point to:

<MIRROR HREF="http://schools.sys.uea.ac.uk/schoolnet/jg/"

the 2nd:

<MIRROR HREF="http://users.aol.com/cnsweb1/"

Any files within these directories and of course directories within 
directories should contain the exact same tag. The browser would have 
to lookup the mirror, then transform the local URL into the mirror one, 
e.g.

[Filepath = http://users.aol.com/cnsweb1/alart/index.htm"]

<MIRROR HREF="http://schools.sys.uea.ac.uk/schoolnet/jg/">

the browser wuld look up the above address and then append 
"alart/index.htm" onto the end.

Additionally, the MIRROR tag could have the ALTHREF element for a 
second mirror, a third mirror could be a second ALTHREF element, and so 
on in the order in which they are read.

All URLs (e.g. .gifs, .jpgs, etc.) could be handled in the same way. 
Certain other resource types (e.g. gopher, ftp) can also be dealt with.

Also, the MIRROR tag could have options set for these other resource 
types.

For the user agent, an option to view alternative URLs to could 
specified, in which case a menu click (or keyboard equivalent) could 
bring down a list of sources.

Comments and suggestions welcome.

 
Regards,

James Green

Term e-mail: jmkgre@essex.ac.uk   |   Home e-mail: jg@cyberstorm.demon.co.uk
Homepage: http://www.cyberstorm.demon.co.uk