forum SignWriting List Forum
  Message 1541  |  Previous | Next  [ Up Thread ] Message Index
From:  Greg Noel
Date:  Wed Jun 30, 1999  1:29 am
Subject:  Re: SGML v. SignWriting

At 10:27 PM +0100 6/28/99, Trevor Jenkins wrote:
>To bring this HTML/SGML/XML thread back on topic if anyone wants help with
>a SignWriter (or other Sign Language) document type definition then let me
>know. I have been an on/off member of the ISO committee responsible for
>SGML and as pentenance for foisting it upon the world I'll make it up by
>helping Deaf users.

I've been thinking about this, too. My hobby horse is making it possible
to display SignWriting easily and efficiently. A DTD (or whatever) to
allow signs to be drawn on web pages without having to download a largish
GIF for every sign would be a big benefit to the community.

It's my belief that you can easily encode a typical sign in less than
twenty bytes, making the transmission of SignWriting only about a factor of
three less efficient than English. (This is not bad; English is a very
efficient language.) Denser encodings are also possible and would probably
cut the size in half, making SignWriting about as efficient as, say, French.

I've been looking at MathML as a starting point---if you look sufficiently
cross-eyed at what they need to do, it's a lot like SignWriting: a set of
resizable glyphs that need to be laid out in a two-dimentional area. They
are designing a plugin that converts the input into a suitable frameset
that could then be displayed by an XML browser.

MathML has a much nastier input problem in that they must parse a
moderately complex language and construct the layout for each equation on
the fly. This is something SignWriting would do in advance so a fairly
simple decoder could determine the layout and pass it to their display
logic to create the frameset.

Since they also want vertical stacks of equations to look "nice" (whatever
that may mean), so they also have issues similar to SignWriting in laying
out a string of composite symbols.

They even have some UniCode space allocated for glyphs, like the integral
sign, which aren't quite like the glyphs that can be taken from other
alphabets. They need a technology to be able to scale the glyphs over wide
ranges, so they're investigating things like TrueType. Those topics have
also shown up on this mailing list.

The MathML effort is very young, which is both good and bad. On the one
hand, nothing is cast in concrete yet, so we might be able to influence
them to shape their design to make it easier for us (that is, chose a
modularization that allows us to re-use their code). On the other hand,
there's no code at all, so it could be quicker just to do our own.
Ideally, the two efforts may be able to colaborate and get some synergy.

So count me in as well. I don't know how much I'll be able to help,
particularly over the next couple of months, but please keep me informed.

Hope this helps,
-- Greg Noel, retired UNIX guru

  Message 1541  |  Previous | Next  [ Up Thread ] Message Index