Re: RFC: Multi-Owner Maintenance robot (MOMspider)

Roy T. Fielding (
Tue, 07 Dec 1993 07:10:59 -0800

>> & BASE? & OWNER? & EXPIRES? )>
>> <!ELEMENT OWNER - - CDATA -- Alias name for document owner -->
>> <!ELEMENT EXPIRES - - CDATA -- Expiration date in RFC850 format -->
> Your suggestion for a multi-owner maintenance robot (MOMspider) seems
> a good one, and I like the idea for making the owner and expiry date
> part of the document HEAD. I never did like the <LINK REL="MADE" HREF=...">
> notation. Perhaps we ought to stipulate the OWNER should be an email address?

I believe that owner/author privacy will become an important issue
as large-scale information resources are added to the Web. Therefore,
I prefer to use a level of indirection such that the owner's alias name
can be used (by MOMspider or other scripts) to look-up the real owner's
e-mail address(es) and perform actions tailored to that owner. For instance,
a {htbin-post | cgi-bin}/mail_owner script could be written which examines
a table of author aliases at that site and determines both whether or not the
owner wants to receive e-mail and what the true e-mail address is.

This also provides a simple mechanism for a single real owner to be
represented by multiple aliases. For instance, this would be useful for
webmasters who are responsible for multiple, semi-independent webs.
It also makes it easier for ownership to change hands.

> When viewing a document with markup errors, or broken links, who should
> the reader mail - the author or the admin staff, e.g. www-admin@host?
> The author may want responsibility, or the document may have been produced
> automatically. Perhaps we should use an additional attribute, what do you
> think?

That is why I used the term "owner" rather than author -- I assumed the
owner would be the one with administrative responsibility for that
document. If the owner wants such e-mail to be directed to some other
admin type, then they would simply specify that admin-type's e-mail
address in the look-up table for the mail_owner script.

> I will be happy to include this in the revised HTML+ spec - so long
> as there isn't vociferous feelings against the idea on the net.

In testing this this morning, I discovered that using a CDATA type
causes existing clients like Mosaic to treat the text between the ignored
<OWNER> and </OWNER> tags as actual printed text -- this is very bad
for backwards compatibility. Therefore, perhaps it would be better
to define the OWNER and EXPIRES elements as EMPTY with argument lists.
That would also allow us to add options in the future.

....Roy Fielding ICS Grad Student, University of California, Irvine USA