how should remote path names be handled?

Thomas A. Fine ([email protected])
Thu, 22 Apr 93 17:05:44 -0400

I was under the impression that when a relative or full path name
at a remote site was found, it was supposed to be converted (by the
browser) into a full http path, based on the current site and document.

Someone mentioned though that this is not the behaviour of some browsers.
So what should we expect? Also, is there a difference in interpreting
a URL at a remote site between file:/blah and /blah ?

Examples:

current document
http://mars.cs.state.edu:80/docs/cartoons/flinstones.html

link found
HREF=jetsons.html
is translated as
http://mars.cs.state.edu:80/docs/cartoons/jetsons.html

link found
HREF=../sitcoms/threes_company.html
is translated as
http://mars.cs.state.edu:80/docs/sitcoms/threes_company.html

link found
HREF=/docs/drama/lalaw.html
is translated as
http://mars.cs.state.edu:80/docs/drama/lalaw.html

Yes, no, maybe?

I think that relative paths should be translated by the browser, and
their use should be encouraged. Full paths should not be handled, and
their use should be discouraged. The reason is the difference in
interpretation between the local and the remote access. Since the
http daemon translates full paths, a full remote path may not point to
a real filename, and likewise a full local path may not be a valid
filename thru the server. On the other hand, relative paths can be
handled correctly both locally and remotely, as long as the relative
path doesn't go outside the domain of what the server will allow.

tom