Re: We need to start talking about group and public annotat

Brian Behlendorf (
Wed, 20 Jul 1994 22:59:28 +0200

On Wed, 20 Jul 1994, Daniel W. Connolly wrote:
> For group and public annotations:
> * If you're interested in a minimum-coding solution, forms
> and CGI-bin scripts are a pretty good hack. HTTP authentication
> isn't that hard -- ask Tony Sanders <>, and he can probably
> point you toward the relavent perl code.

The trick is to get something that scales well. Given that we want
someone who calls themselves "TimLeary" when they post a comment to our
site to keep and protect that name, we have to build a system that can
handle hundreds, thousands of distinct password-protected names
(recognizing this is only weak authentication). At least, this will be
true until SHTTP gives us true user-authentication. It also needs
to scale well in terms of the number of documents we expect. I've been
warned that nothing less then a commercially-supported object database
will be needed.

> * If you're interested in a scalable solution, have a look
> at USENET news! The only trick is to enhance browsers
> to understand the message/rfc822 content type a little better,
> (including message/rfc822 bodies whose bodies are text/html)
> and to support NNTP posting and thread navigation.
> Conventional newsreaders are bad because they keep a connection to the
> NNTP server open during your whole section. But I'm sure they could be
> modified to use the gopher/http connect-doit-disconnect style. But
> you could have public NNTP servers that zap connections after they
> go idle after a few seconds. Then folks with conventional newsreaders
> would get the hint that they're not welcome.
> You can use the USENET newsgroup hierarchy or make up your own.
> You don't have to expire old articles.

The problem with USENET, as I pointed out before, were the issues of
control and interface... and I'd add now that I'm much more encouraged by
the thought of getting stuff done in the WWW community than in the USENET
community :)