If you're reading this at about the time I wrote it, some of the#

Thu, 31 Oct 2002 17:44:26 +0000

If you're reading this at about the time I wrote it, some of the preceding (lexically following) diary entries are really rather short. Apologies for that. I might go back and fill them in later, but in the meantime: the URL was written down during Duane Rettig's simple-streams tutorial, and should point to Paul Foley's simple-streams implementation for CMUCL. The aardvark was an example in Nick Levine's CommonSQL tutorial.

The conference (I'm at the ILC, if I'd forgotten to mention it before) is now about 80% done. Noteworthy points

• McCLIM rocks

• I met Gary Byers, who tells me that there's a perfectly good reason for the trick with lowtag punning that CMUCL/ppc was doing. I'm going to have to reread it before I understand his point well enough to explain it to anyone, but it's related to (avoiding the pain of) interior pointers into objects

• Roger Corman talked about native threads and how he added them to Corman Lisp. The good news is that we've already been doing some of what he's had to do; the bad news is that without having native threads as a driver to make sure we get it right, we probably haven't been doing it particularly uniformly.

• Open Source vs Not: Kent Pitman gave a short talk in which he managed to assocate the concepts `college students', `don't have to work for a living', and `dumping' with authors of open source software, but without actually at any time directly stating that said authors are or are doing any of the above.

• Metadata: lots of stuff about the semantic web, with exciting pointy-bracket-acronyms (RDF, DAML+OIL, OWL, etc). My opinion continues to be that
  • metadata is just data, and if you don't trust the actal data to be true (we're talking about web pages here, so you certainly shouldn't) why would you want to trust the metadata either, given that it's probably been provided by the same people. I choose to interpret remarks by Peter Norvig (talking about people trying to stuff up Google) as supporting this position

  • highly structured data (all this angle bracket soup) tends to lead to brittle and unadaptable systems, and this doesn't scale. The virtue of the Web is that it works despite being essentially mob-driven. RPG, I think, would agree with that. In fact, RPG said exactly that, just not in the context of metadata.
• Lisp is in decline, because people in the various US Government departments using it are agitating for its removal. Um. This is a different belief than "because of the AI Winter", but holding it still precludes you from adding yourself to the CLiki YoungLispers page. People were saying the same thing about Unix in 1994, I'm sure.

• Lisp is unsuitable for Google because implementations don't cope with their scary scaling requirements. More generally, of the ten or so pretty-much-unique advantages of Lisp ten years ago (GC, interactive development, native support for collections, etc etc) only about two are still unique. Norvig considers the syntax to be one of them (if you're doing macros, it is).

  My comment on this is that you can't yet get all of the other eight advantages in the same language.

• MCL 5.0 got announced

• I appear to have volunteered to help with the ALU web site, and with any luck we might have some progress forthcoming soon (like, next week). The plan is to turn it from a perfectly preserved museum piece into a living site that people actually want to visit more than once.
• Next year's conference (the plan is that there will be one, yes) is probably at about the same time of year, in New York.