ILC2007 was a roaring success in my opinion. Our benevolent organiser Nick Levine is unlikely to ever pay for beer at a Lisp meeting again.

I was extremely lucky to have long conversations with interesting and unexpected people over meals and drinks. Richard Brooksby explained a lot of interesting things on memory management and software construction, Scott McKay on programming in general and some particular programs of interest, Chris Wright on broader computing topics than I'm accustomed to discussing with a single person (not a bad repetoir for a practicing doctor), Jeremy Jones on the development of MCL and life at Xerox PARC -- to name but a few.

Great to see the ECLM crowd again of course - too many to enumerate except perhaps by absence: Thomas, Nikodemus, Tim, Tony.. actually too numerous that way too! I hope we'll see them and people like the Dylan Hackers and Charlotte Herzeel joining the fold at ECLM 2008 in Istanbul. Zach was missed despite our never having met him!

I feel very alive after spending a few days in the same place with so many Lispers doing so many different things. The universe is certainly chugging along in the right direction.

On to more specific impressions:

• Memory management is extremely interesting. The Ravenbrook MPS tutorial gave me the healthy mixed feeling of stupidity and curiosity that I so love.
• CONS has been removed due to bugs in the garbage collector. Please update your programs.
  Franz have some surprising new ideas about stylish Lisp programming for high-performance applications. That to one side, AllegroCache sounds like an excellent program.
• Robert Strandh and McCLIM keep on marching forward!
• ITA Software are employing every Lisper they can find and better than that they're funding people to work on SBCL and even SLIME! I'm very happy to see the Lisp free software industry begin to thrive. Heady times these are before the airline search (AS) winter! :-)
• Peter Herth's presentation of LTK was a happy dose of simple practicality and good taste. Bravo! Would that TK interfaces themselves were much more appealing though.
• CL-HTTP: Mallery would have been eaten alive at an Apache conference!
• Liskell: Syntax is what Haskell got right! Clemens will surely realise this and be a full-blooded Haskell hacker by this time next year. Recall from genuinely enlightened discussion on the ll1 mailing list that metaprogramming is much less important in Haskell than in Common Lisp.
• Apologies once more to Mike Sperber and Didier Virna for revealing a glimpse of the murderous rage lurking deep in my soul when they mentioned being XEmacs maintainers. :-)