A mail to some comrades, seeking advice on how to do 
annotations
in warp.

---------- Forwarded message ----------
Date: Sun, 24 Mar 2002 19:29:46 -0500 (EST)
From: cdent@burningchrome.com
To: XXXXX, XXXXX
Subject: warp, purple, annotations

Hi, I'm having some ideas about how to modify my warp software,
and I need somebody to bounce them off, so I thought of you two.
Hope you don't mind. I considered posting this to the ohs list
directly, but my ideas seem a bit flaky and my awareness of
existing information is slim, so I thought I'd start in a more
confined space.

Warp remains something of a toy for me, but its proven to be an
interesting place to play out some ideas.

The basic idea at this point is wanting a way to associate
commentary with the entries (definitions) in the database in a
granular fashion.

What I'd like to do is purple the definitions as they are saved,
each chunk gets a purple number of some kind. Then when the definition
is presented each chunk shows its purple number, in a style similar
to the way purple numbers are done in the bootstrap archives.

Also associated at that location would be a small icon for adding
a comment. Clicking there would bring up a form for adding a
comment.

When the definition is next presented, the comments at each
purple number would be visible in some way. I'm not sure how at
this point, maybe by inlining them with the text, or doing more
floating java script stuff, or # type links to something further
down the page. That's not the issue at this point.

What's a real issue, and seems to be the issue for much of the
discussion on the ohs list, is the following question:

  What happens to the comments when the word is updated to have a
  new definition? If the new definition is comletely new text, it
  doesn't matter, the associations are gone anyway (but are still
  maintained with the archived, and available, older defition).
  But what if some or most of the text is maintained. How to maintain
  the association, and the purple numbering?

This seems quite the conundrum, especially since one of the
design goals with Warp is to keep things cheap and simple.

I've considered doing some kind of statistical diff, but that
strikes me as _hard_.

I hope I've explained this well enough to be clear. Let me know
if not.

My warp is at (I'm up to 5 of them now, as I've had people ask
for their own):

  http://
www.burningchrome.com/~cdent/

If you want to play around

  http://www.burningchrome.com/~cdent/globalwarp/

is open for anyone to change.

The data is strucutred in a two table database:

  - definitions:
      defid  (pk)
      wordid (fk from words)
      definition
      modtime

  - words:
      wordid (pk)
      word

One to many relationship between words and definitions.
Annotations would be an additional entity with a one to many
relationship from definitions, with a further modifier indicating
the "purple place".

I think. That's a bit grey so far.

Anyway, I appreciate any comments you may have, especially on the
issues with migrating comments and purple numbers from definition
to definition (within the same word). I believe there have been
some references to resources on the list, but they are hard to
find (which is something I'm working with Kathryn La Barre to see
if we can find some ways to deal with that, but that's a separate
conversation), if you guys remember and could point them out,
that would be great too.

Thank you very much.