Grex Info Conference

View Responses

Grex Info Item 309: About URL
Entered by arasu on Mon Jun 15 08:58:00 UTC 1998:

Please explain about the URL:

19 responses total.

#1 of 19 by valerie on Tue Jun 16 03:30:09 1998:

This response has been erased.

#2 of 19 by srw on Mon Jun 22 21:45:27 1998:

A URL is made up of several parts. The first part is the protocol. Let's 
take a look at Grex's homepage URL - http://www.cyberspace.org/

the http: portion is the protocol. There are many other protocols 
besides http, but it is now the most common by far, so that modern 
browsers all supply it if you leave it out. However, if a URL is given 
in a document as a link without a protocol portion, the protocol is 
construed to be the same as that used to access the document itself.

Other protocols you may have heard of are ftp: and mailto: , but there 
are quite a few others, too.

The //www.cyberspace.org/ portion is the host name portion of the URL. 
It identifies the machine whcih contains the document that the URL 
refers to. It must be looked up in DNS server unless it is given in its 
numeric form. Numeric host IP addresses will work but are not 
recommended, because they can change unexpectedly, rendering your URL 
useless.

If the host name is not given in a link, the hostname of the host 
containing the pointer is supplied. This is the best way to create links 
internal to a website, as the whole site can then be moved to another 
host without having to edit all of its embedded URLS. 

There is no more after that in the URL above, but URLS may contain 
(optionally) two further portions. The first is the path of the document 
in the directory tree of the server. This portion looks like

/foo/bar/baz/filename.html

If a URL is specified with no host and a path that doesn't start with a 
slash, the path is taken relative to the directory containing the 
document in which the pointer appears.

The second portion is an argument. It is delimited by a ? and signifies 
that all that follows is an argument to be passed to the named document. 
The named docuemnt in this case is usually a program, and not a static 
page. You can easily see these in a backtalk URL, for example.

#3 of 19 by davel on Tue Jun 23 09:50:22 1998:

> ... Numeric host IP addresses will work but are not
>  recommended, because they can change unexpectedly, rendering your URL
>  useless.

Heh.  Of course, that *never* happens with non-numeric IP addresses ...

#4 of 19 by mdw on Wed Jun 24 22:26:02 1998:

URL's can also contain a # character, for tags.

#5 of 19 by remmers on Thu Jun 25 11:32:27 1998:

Although if memory serves, the result is called a "URL fragment".

#6 of 19 by valerie on Sat Jul 11 13:24:30 1998:

This response has been erased.

#7 of 19 by scg on Sat Jul 11 15:57:32 1998:

It happens less often with fully qualified domain names instead of IP
addresses.  If a domain name points at the wrong place, it's most likely a
screwup, which will hopefully be fixed quickly if the person running the site
cares, whereas IP addresses tend to change intentionally.

Also, a nitpick to Valerie's nitpick:  mailto: is part of html, not http. 
http is a protocol used for transferring data, and doesn't particularly care
what's in the data.

#8 of 19 by remmers on Sat Jul 11 19:44:40 1998:

And here is a nitpick to Steve's nitpick to Valerie's nitpick:
mailto: is part of the URL syntax as defined in RFC 1738. This is
independent of both HTML and HTTP, although it is used in HTML
(and in other standards as well).

#9 of 19 by rcurl on Sun Jul 12 04:54:03 1998:

(I love watching the gurus nitpick...its like watching chimpanses.....)

#10 of 19 by srw on Sun Jul 12 05:40:10 1998:

<runs and hides until it is over>

#11 of 19 by albaugh on Sun Jul 12 08:03:51 1998:

(And keep in mind that chimps are both meat-eating predators & canibals)

#12 of 19 by rcurl on Sun Jul 12 16:42:26 1998:

What I had in mind was their picking nits off each others' heads.

#13 of 19 by mneme on Sun Jul 26 04:13:59 1998:

Actually, strictly speaking, mailto isn't limited to being part of the HTML
protocol either -- it's a URL type, one of those allowed by many versions 
of HTML, but when you get down to it, as far as HTML is concenered, it's
just a protocol, not part of the markup.  It's Certainly not part of HTTP,
though; an entirely different protocol.

#14 of 19 by remmers on Sun Jul 26 12:38:38 1998:

Isn't that exactly what I said in #8?

#15 of 19 by davel on Sun Jul 26 18:53:29 1998:

More or less.

#16 of 19 by curby on Thu Jul 30 07:10:56 1998:

Not really a nit, but more of a clarification of John's nitpick to
Steve's nitpick to Valerie's nitpick.  While RFC 1738 first defined the
mailto URL, proposed standard RFC 2368 is the most recent rendition of
the mailto URL and looks to make it into full RFC status relatively
shortly.

        RFC 2368:

        Title:      The mailto URL scheme
        Author(s):  P. Hoffman, L. Masinter, J. Zawinski
        Status:     Proposed Standard
        Date:       July 1998
        Mailbox:    phoffman@imc.org, masinter@parc.xerox.com,
                    jwz@netscape.com
        Pages:      10
        Characters: 16502
        Updates:    1738, 1808

        URL:        ftp://ftp.isi.edu/in-notes/rfc2368.txt


Of course, maybe this is more then you needed to know...

#17 of 19 by remmers on Thu Jul 30 17:51:13 1998:

Thanks for the clarification. I just can't seem to keep up with
all these RFC's...

#18 of 19 by jared on Fri Feb 19 18:35:06 1999:

be sure to check out the rfc-index

#19 of 19 by myindia on Fri Oct 15 17:08:25 1999:

I am not much sure , but I think that if "mailto" would not have been
a protocol then one would not have been able to use "://" directive after
"mailto"

Response not possible - You must register and login before posting.