Re: relative URLs VRML viewed from HTML-browsers

Brian Behlendorf ([email protected])
Fri, 12 May 1995 19:31:46 -0700 (PDT)

On Fri, 12 May 1995, David Mott wrote:
> > If a browser for a file format is going to 1) talk HTTP 2)
> > contain relative URL's and 3) be designed to run stand-alone, the concept
> > of a <BASE> is necessary.
>
> You had to load a file in initially, right? The URL you gave
> to WebSpace to get it running is that BASE URL from which
> relative links are resolved.

Well, that's true of HTML as well, yet people still found a need for BASE
in HTML even in the first draft.

Consider: a collection of HTML documents on a server that has a full text
search engine. When the search engine finds matches, it links those hits
to cgi scripts that highlight the looked-for words in the document. Thus,
the document has a URL when returned like

/cgi-bin/return-doc/dir/cleveland.html?word=Cleveland

If there is a relative link to "cleveland.gif" in the HTML document, the
server will try and fetch

/cgi-bin/return-doc/dir/cleveland.gif

which would be incorrect. If I had a <BASE> tag in that document, all
would be well - in fact, I could even have the return-doc program insert
a BASE tag for me.

I'll now concede, however, that this could wait until 1.1.

Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
[email protected] [email protected] http://www.[hyperreal,organic].com/