On 1/11/2000 at 3:56 PM [EMAIL PROTECTED] wrote:
{
If someone sends me a description/specs of 1., I can certainly respond
with some ideas or - more likely - alpha quality scripts.
}
Elan,
Here are some design notes about my "Kabob" project, which Bob Racko
has pointed is much like the various Wiki applications. I'm starting to
work with Wiki myself. Maybe we can just start calling this Wiki/REBOL.
As stated, the end-game is something like the dmoz.org directory that
leads to hypertext documents rather than Web sites. They have really
done an amazing job over there of putting together a keen suite
Web-based groupware tools. Tres kewl.
Anyway, if a person wanted to add a page to the directory, they could
select a link to bring up a form with the appropriate fields. If the
directory is a knowledge base about REBOL, you might be able to go to
the 'Series heading and post a document there about immutable series.
Ideally, people should also be able to post quick "talk back" replies
to your document, and have those listed. They may also fill out a quick
form to have the document emailed to themselves or a friend. Someone
else might post a related article, and add a link between them ("see
also"). You might come back later and update the page, and also a
comment to the change history. Anyone who was interested might also
have subscribed to the page, and would recieve the updated copy by
email. They may also be interested in other documents you may have
posted, and could lookup your author profile for links to your other
articles.
Of course, the page could just as easily be a FAQ, a newspage, bug
report, script, list digest, or anything else.
None of this is new ground. There are examples of Web pages that do all
of these things, including some REBOL scripts in our library. I'm just
proposing we setup an easy way to create, updated, and organize such
pages into a structured knowledge base.
Now we probably have a good vision of how this should look in action.
But how is it going to work "under the hood"? (or "bonnet" for those
using the UK dialect:-)
The obvious thing would be to leverage the server's file system, as
dmoz apparently has, and create folders and sub-folders to represent a
hierarchy. You could then just add a document to the appropriate
folder, and perhaps make use of the server's security system.
Though, there may be advantages in storing the documents together. For
example, they could be stored equally under separate headings, and it
might also be easier to replicate between sites. It would also be more
portable, which I think is part of being REBOL.
Likewise, the obvious thing for the document directory would be to make
it a database. Though, it occured to me that the directory pages could
also be documents. So when you add a document to the directory, you are
really updating the link list on another document.
Of course, making all this available in a database format would be an
interesting idea. My thinking would be to put enough information in
each page (perhaps hidden in comments) so that it could be read into a
database if needed. (Or could reconstruct the database if it crashed.)
Essentially, each document could act like a database record, with the
Web folder binding them together into a database. Then, if desired, we
could also read in the documents and create a monolithic file.
Accordingly, we might want to assign the documents key IDs as file
names, and, in general, just display document titles to humans.
Ok, enough theory, where to begin?
The first core work could be simply a browser-form to HTML-page script,
like Grossman's Wiki. It should also store the article title, meta
keywords, original author, creation date, last editor, edit date, and
edit history.
To complete the first usable system, we might setup up a script to
create an updated index by scanning the document folder. Ideally, there
should be at least chronological (today to yesterday) and alphabetical
indexes, along with a list of documents by author.
At this point, we'd have a bare-bones, automatic submission system,
ready to store plain HTML documents in a single "shoe box".
Of course, we don't want to be writing in stone, so we would next need
+ A HTML-page to browser-form script for updating.
+ A format for submitting and updating by email.
+ A secure registration system for authors and editors.
Which brings us to a more useful multi-authoring system.
The next big step would be creating a hierarchical index ...
(Incidentally, Andrew Martin is busily at work with his HTML dialect,
which might have a place here too. We wouldn't need much HTML to get
started, but may want more later, so we might want to build this in
from the beginning.)
-Ted.
