Monthly Archives: April 2005

Gardner museum

On advice from Rob Harris, I went to the Isabella Stewart Gardner Museum. Unfortunately, I only had about an hour to visit, since the musuem doesn’t open until 11:00 am, and I had a flight out in mid-afternoon. Still, it was a wonderful hour, well spent enjoying the gestalt of the musuem, famously controlled by Gardner’s will to neither add nor subtract from the musuem’s holdings. The courtyard is delightful; it, and the sight of a stained glass window as I turned a corner, both made me gasp.

(Err, corrected to’s tagline is “Do You Need Parent Rolls of Tissue?”)

Mary Mac

I’ve been in California for work, and I got to play a bit too–I sang at the first annual San Francisco Sacred Harp singing on Saturday, and I went to Yoshi’s jazz club in the evening. At the Sacred Harp convention I met Mary McDonald-Lewis, who came down from Washington state to sing. It turns out I’ve heard her voice before–she’s the voice of GM’s OnStar in-car voice recognition system. There’s a great fun MP3 of her demoing her voice on her website at

Apex in tha house

Apex in tha house (created via HTTP in tha House).

variety of task domains including those listed ask developed by users combines components yes developed by users outside on cognitive architectures qi a more general airspace definition language pdl see

were developed by users outside building debugging analyzing knee manual for more information what apex vary greatly their experience uncut explore autonomy design parameters simulating for teaching and te capabilities for monitoring and sequence of routine free


(Apex is the Autonomy architecture that I hack on.)

Multiple values in binary trees

I had a small ‘ah-hah’ experience while reading Implementing sets efficiently in a functional language. Descriptions of binary tree sorting/searching algorithms, such as the excellent Introduction to Algorithms, state that items to be sorted have to be strictly ordered, which means you can’t have duplicates in the tree. But this is easily solved by storing duplicates together in the same binary tree node. “The implementing sets” article suggests adding a ‘count’ field to the node, which marks how many copies are in the node. Often, though, the key and value of a node are different–which suggests that (if the values are strictly ordered as well) you can use the same kind of binary tree to store values. Often it is the insertion order that matters, in which case this can be used as a key.