J.P.Lewis just made my day. He sent me a pointer to his new project VisualIDs. Beautiful stuff!

This paper points out that for the spatial desktop metaphor to work as intended, it really needs scenery (distinctive appearance) in addition to the spatial data layout that we have now. The paper shows how suitable appearance can be automatically generated (using a scheme algorithm). User studies show dramatic improvment in file finding after only brief use.

I mentioned his Mathematical Limits of Software Estimation paper in February.

Colophon

The manuscript of The Baroque Cycle was written by hand on 100% cotton paper using three different fountain pens: a Waterman Gentleman, a Rotring, and a Jorg Hysek. It was then transcribed, edited, formatted and printed using emacs and TeX. When it was totally finished, the TeX version of of the ms. was converted to Quark XPress format using an emacs LISP program written by the author. Some share of credit thus goes to the people who made the GNU/Linux operating system and to the originators of LISP.  Maps were produced by Nick Springer with useful input from Lisa Gold, who also organized the family trees and assisted in the preparation of the Dramatis Personae. The geometrical illustrations (Apollonius of Perga's conic sections and the woodcuts from Newton's Principia Mathematica) were prepared by Alvy Ray Smith, working from scans or photographs of old books.

Update: [Link] (via David Magda)  

Way, way back in 1983, when I decided to learn 6502 Assembler (yummi!) and Apple Pascal (retch!), I borrowed a xeroxed copy of p-Source: A Guide to the Apple Pascal System by Randall Hyde, which had as it's cover a funky adaption of American Gothic by Grant Wood. Randy is an exceptionally talented writer. I just received a used copy of p-Source and after all these years it's still amazing. The p-code interpreter was the first virtual machine I dissected, debugged and enhanced. Randy writes about the implementation in section 3, chapter 1 "Modifying the Apple Pascal P-code Interpreter":

As it turns out, the students who originally wrote the 6502 p-code interpreter were fairly inexperienced on the 6502 microprocessor chip. As a result of this inexperience, the p-code interpreter is not as optimal as it could be. If the 6502 interpreter were completely rewritten using better coding techniques an overall increase of 15-20% could be realized. That's almost as good as the 6809!

Randy also is the author of The Art of Assembly Language Programming and the soon to be published Write Great Code: Understanding the Machine. Randy also wrote an inspiring article about Why Learning Assembly Language Is Still a Good Idea.

Why is this still relevant? Some might think it's a crazy idea that the DISASSEMBLE function ever made it into the language standard. I disagree. As Robert Pirsig tried to explain in his seminal book Zen and the Art of Motorcycle Maintenance: An Inquiry into Values, only in understanding the inner workings of a machine can we start to understand the nature of quality. Because Pirsig wasn't a programmer himself, he couldn't just write something like: "The Vulcan mind-meld with the machine is at the center of good code."

The availability of DISASSEMBLE in the Common Lisp standard is a message: There shouldn't be anything between you and your machine. When driving a TVR or programming a Lisp Machine, you want to be in control.

We experienced some server problems since noon yesterday. The blog is working again, but I'm not sure if the problem is really fixed.

Last weekend my best friend Peter finally married Natascha. Somehow I'm sure this two day event will be remembered as the mother of all weddings by all participants.

The wedding took place in Weinheim, a beautiful city with a 2800 year history. After church we followed the white Cadillac convertible to the restaurant Pfeffer und Salz (Pepper and Salt), owned by the father of the bride, where we had a spectacular party drinking fine wine and champagne, eating delicious food and dancing all night long.

yrk quotes an article titled Scheme vs. Common Lisp at Philip Greenspun's site. Delicious!

Finally, Gerry Sussman said "Of course, you can't expect Lisp to do something like that; Lisp can't do things like this. If you want to deal with massive data sets, you have to use C. It is sad but true." I said, "Gerry, I think this would run on an old 1 MIPS 3600 with 8 Mbytes of RAM. I'd just rewrite this in Common Lisp. The code would be cleaner because of the generic functions for sequences and I'll throw out my private sort function and just use the one built into Common Lisp." Sussman said "If that is true then the whole Scheme project has been a waste and I'll shut it down."

I'm happy to report that, due to Martin Becker's relentless efforts as our editor , the print edition of Successful Lisp is nearing completion!

We'll be reviewing the completed drafts in a couple of days and I will be able to make sample chapters available online.