PLI 2002 -- report from someone who was there?

[Ulf Wiger]

On Wed, 23 Oct 2002, Matthias Lang wrote:

>There are a bunch potentially interesting papers there. I
>can't find an online copy of the proceedings anywhere.
>Anyone know? Anyone who was there care to comment on how
>big the workshop was (how many people)?

The proceedings will hopefully appear sometime through ACM.
At least that's the general idea.

The workshop was quite good, even though the turnout was not
exactly enormous (about 30 people, I think). I heard a
rumour that some other workshops had bigger problems.

I did peek into another lecture hall where lots of people
sat listening to someone pontificating about software reuse.
Oh well. Software reuse was covered summarily in our
workshop by (I think) Martin Logan who stated that there's
lots of it in Erlang/OTP. (-:

Short summary:

- Phil Wadler explained to us that God gave us static
  typing, but apparently not Erlang. However, God wants
  us to use functional languages, so we're probably sort of
  OK anyway. Phil also expressed great pride in being
  invited speaker at an Erlang workshop, since Erlang is
  _the_ most successful functional language in the world,
  and the one with the most millionaires (no one cared to
  explain Swedish tax laws to Phil -- God didn't give them
  to us, anyway.) Lots of theory, and lots of T-shirts
  coming off... well, you just had to be there.

- Richard Carlsson talked about the new package concept.
  It's coming together, but I've come up with some more
  issues, to which I will return later. Nice talk.

- Kostis explained about how Hipe may double the speed
  of Bit Syntax matching. Promising.

- Thomas Arts gave a high-speed practical demo of his
  trace analysis program. Great stuff, I think, for
  those who want to visualize their Erlang programs
  in different ways.

- I talked about testing with Erlang, and to my surprise,
  the auditorium didn't fall asleep. Must mean that people
  are actually using Erlang for some real stuff.  (:

- Scott Fritchie talked about the Erlang Driver Writer's
  Toolkit. This is really good stuff. Kenneth Lundin may
  have to stop telling people that it's unsafe to use
  linked-in drivers...

- Hal Snyder and Martin Logan talked about Erlang use at
  Vail Systems. Same old story about how a small group
  of people are allowed to use Erlang to monitor and
  restart all those Java apps that can't stay up by
  themselves -- but have to fight to be allowed to tackle
  the really interesting stuff. Just kidding. Good
  presentation, down-to-earth, factual, success story.

- Thomas Arts again, describing how they analysed some
  properties about a Video on Demand system, using
  verification techniques on actual Erlang code to
  determine dimensioning properties like number and
  size of disks, distribution of movies across disks,
  bandwith, etc. State of the art stuff, indeed.

- Kostis offered some ideas on how to speed up
  inter-process communication in Erlang. The paper
  states the goal as being "to have truly lightweight
  processes where message passing is at least as
  efficient as method invocation in a modern object
  oriented language". Some interesting ideas, but
  nothing firm yet. Hope they succeed.

- Joe Armstrong talked about how his protocol description
  and contract definition syntax does everything XML
  and WSDL does, and more, but much more beautifully.
  Always fun to listen to Joe, and I've already started
  playing with his new toy. Very interesting.
  http://www.sics.se/~joe/ubf/site/home.html

- Closing comments, and a preliminary report on the
  questionnaire. Everyone seemed happy.

/Uffe
-- 
Ulf Wiger, Senior Specialist,
   / / /   Architecture & Design of Carrier-Class Software
  / / /    Strategic Product & System Management
 / / /     Ericsson Telecom AB, ATM Multiservice Networks