Hi Joseph,

 > At the weekend, I gave a short (20 minute) talk on LaTeX3 to the UK TeX

you must be a fast talker: 37 slides in 20 minutes ...

 > User Group Speaker Meeting. I've put the slides I used on my website
 > (http://www.morningstar2.demon.co.uk/talks/LaTeX3-from-outside.pdf) so
 > that people can see what I discussed.

I think this is a nice outside view on what is there right now. do you mind if
we add this to the latex project web site with some suitable title to get it
to some broader audience?

 > Broadly, I covered two areas: what I see as the current ideas
 > implemented in LaTeX3, and what I've been doing to experiment with the
 > new syntax.  Hopefully I've done a reasonable job on both parts!

i think so

 > Will Robertson suggested that the feedback I got from the talk might be
 > useful here.  A few things were discussed, and I've tried to summarise
 > the key points below.
 > 
 > There were a few quite technical questions. One question was how the new
 > kernel would affect the ability to create a "secure" LaTeX. I suggested
 > it would make life easier, but couldn't be sure that LaTeX3 would
 > definitely be suitable for creating a secure version. I was asked about
 > thread safety, and said that the engine was the obvious limit there. The
 > idea of "object orientated LaTeX" was raised. I didn't discuss whether
 > this was desirable (I'm not sure myself), but again I said that the
 > engine (even with Lua) would be a limitation.

my take on "secure" and OO-LaTeX is that the former is an illusion (unless you
take away the power of a self-modifying language on "all" levels ... and I for
one don't like to do that) and OO-LaTeX seems to me more a question of
desiging an OO-Formatter a language on a TeX based formatter (and that is what
LaTeX is attempting isn't  ever going to be successfully OO

 > There was some discussion about mixing existing TeX primitives/code with
 > LaTeX3. I said that at the moment this works, but may not in the future.

my take on that is yes, there will be a point in the future where we have to
build a kernel without any 2e support inside (that will be the time where we
have to provide all major functionality as part of l3 packages or else this
will fail ...)

 > The comparison with the MathTran secure TeX daemon was made, as it makes
 > many primitives undefined. There was some confusion about the current
 > need to run LaTeX3 "on top" of LaTeX2e, although this was resolved (a
 > comparison was made to running Windows Vista on DOS!). 

for any one interest there was a full l3 kernel not containing any 2e code or
support (based on a far earlier version on expl3). since then we developed the
language a lot further but at the same time taken out some of the more arcane
implementations to make things a full kernel.

right now expl3 is on a 2e kernel for 2 reasons 

  a) because essential parts that for a full typesetting language havn't been
  done in a way we want it (more or less all the higher level stuff) and

  b) we though that experimenting and writing code that could integrate with
  the old world was beneficial at this point in time

but it has to change at some point

 > I mentioned that as I see it when a LaTeX3 format is available most
 > existing packages will not work (assuming a new kernel is written entirely
 > in the new syntax).

true. as i said: while this step will essentially be necessary at some point
it will only work if we manage to rebuild all the major functionality in l3
packages first (or at learly the same time), basically the amount of stuff
described in TLC2 (even though not necessary the exact packages or the actual
syntax they use)

 > I also had some discussion about where the new code is in terms of being
 > ready to create a new format. I said that the current coding side is
 > more-or-less complete, but that I think the document design and user
 > side has a long way to go. I also mentioned that LaTeX3 will (I imagine)
 > need to cover a lot of what I currently done in separate packages. That
 > seems to be a long way from happening (for example, the basic classes
 > need to have the flexibility of something like memoir). 

yes page design and document design, memoir, geometry, ... you name them


 > On the
 > "readiness" side, I also pointed out that there are still a few rough
 > edges (for example, \etex_scantokens:D and the l3messages module).

i respectfully disagree: l3messages is just leftover rubbish ... and we should
simply do some drawing board exercise and design something suitable (i think i
said this before)

 > There was also some comment on the length of time things have taken
 > (l3messages came up again, as an example of where this shows up). I said
 > that things do seem to have picked up recently. I said that I was
 > hopeful that we'll see real progress (for example, finalising expl3 so
 > it can be used more generally) soon, but again that is down to the team.

i think we are on a good way doing this (though right now the fact that
company I work for got bought didn't quite improve my time availability) but
going over the expl3 language as we did recently helped us collecting a lot of
smaller things that should be fixed to get the language straightened out
further. that (probably last) bigger revision should then put the more
technical part of the language in a state that we concentrate again on things
like the output routine and the high-level functions 

 > I hope this feedback is useful to the team and other interested in the
 > future of LaTeX.

indeed. thanks. and sorry for being fairly uncommunicative in the recent past
(and probably bit into the future)

thanks
frank