Re: Why is From: limited?

>Chuck Shotton writes:
>> Suppose "bin" owns all WWW documents? Do you think a system admin
>> wants to spend all day forwarding stuff to the appropriate person?
>Wouldn't any sysop who is so misorganized to install the web pages in
>the bin account and generate From headers from it deserve this?

The arrogance (or ignorance) contained in this statement is amazing! Why
should an applications level communications protocol like HTTP have ANY
impact on the way I choose to administer ownership of documents on my WWW
server? You Unix guys keep getting lost in implementation details that are
hacks and are not cross-platform solutions. Cut it out! Using "user and
group" info from a specific O/S as a mechanism for identifying authorship
of WWW documents is about the most fragile, useless way I could think of to
do this.

Some sort of database scheme where HTTP entities are tagged with attributes
like owner, author, expiration date, etc. is the ideal mechanism for
generating all of these meta headers that don't (properly) come from the
file system.

>> I didn't say it might not be a good idea. I said that it couldn't be
>> implemented in an efficient, usable fashion across multiple
>> platforms.
>If the HTTP servers running on Microsoft can provide rev=made links
>they can also provide From headers. If not, then not.


>> HTTP servers ultimately need to sit on databases (object bases), not
>> file systems. When this happens, things like this will be easy to
>> implement.
>And then what header will you use?
>        Link: <>; rev="Made"; title="Tim Berners-Lee"
>        From: Tim Berners-Lee <>

What difference does it make? At that point, you actually have reliable
information that means what it says, and not some ambiguous or inaccurate
data pulled out of the file attributes of the document.

>Please don't tell me the From is so much harder to implement than the
>Link or more compatible with existing software on the Internet.

That's not the point. The point is that both are equally hard to implement
in a reliable, accurate fashion. You cannot ever convince me that using the
file ownership attributes to determine this info, even if every Web server
on the planet runs on Unix, will ever be 100% correct. Your premise is
based on the faulty assumption that Web servers are built to serve files
out of file systems, and this couldn't be further from the truth. And this
is only going to become more obvious as WWW software moves from the realm
of hackers into the realm of real-world programming, where things are
designed and engineered or they sit on the store shelf and rot. These
servers will serve exclusively from content residing in databases. File
system based servers are definitely an endangered species.

>> In the meantime, the value of such protocol candy is minimal when
>> other, better techniques exist in HTML for doing the same thing.
>The HTML tags have to be added manually and that's why so many web
>documents lack authorship information.  An automatically inserted
>optional header could help in those cases and may evolve to become the
>preferred tag because it's added automatically like the headers of my

In existing server implementations, yes, such tags would have to be added
manually. If it is important to you for people to know you are the author
of a document, I suggest you add them. For others, they may not want to be
pestered by reams of e-mail just because they happened to put a document on
the Web. You are completely overlooking the need for anonymity in your
blanket implementation.

>> WWW is for anonymous publisher to anonymous reader communication.
>Like news, WWW is a communication from prominent publishers to
>anonymous readers.

This is your opinion. How often to you bother to check when following links
around the Web to see if you've wandered off to a different site without
realizing it? The very nature of the Web encourages seamless integration of
content. This, more than anything, makes knowing who the publisher is at
any given time difficult. And some people actually count on this.

>> Please read through the URI/URN standards and the HTML 3.0 standard
>I did.  From is not in there yet.

Geez. No, but a way to specify document ownership/authorship is. Did you
read that part, or did you just grep for "from"?

>> you'll find plenty of stuff to sate your appetite for features.
>I am not asking for new features, I am only trying to prevent an old
>and well-established and simple standard from being forgotten.

As long as it is implemented your way...

Chuck Shotton                                                   "I am NOT here."

Received on Saturday, 1 April 1995 07:01:21 UTC