August 2009 Meeting Notes

Compiled by Dave Jaffe

Contributions from Kevin Appert and others

10:00

Coffee and a Chat

10:15

Videos from July 2008 Continued
That month our room at Cogswell wasn't available, so a few members convened at Ting's meeting house in San Mateo. Masa videotaped most of the meeting. Was Bill Ragsdale there? Dave Jaffe was there, so we'll have a look at:

Interesting Items - Volume 11 - Dave Jaffe
Dave continued his survey of interesting newly-introduced microcontrollers, I/O peripherals, sensors, products, and gadgets that can provide useful and unique capabilities to Forth embedded systems.

In the morning, we ran the tedious video of the July 2008 meeting. It included a rambling discussion among those present. The interesting part was Dave Jaffe's "Interesting Things" presentation which was interrupted in the middle when we ran out of time. Dave's slides can be viewed here.

11:40

Lunch
Lunch on-campus is most likely, but open to discussion. To avoid the "noon lemming effect", we'll leave a little early.

13:00

Introductions, Announcements, Rumors, Random Access

We watched:

Hardware Wars

What the Hell is an Aluminum Falcon?

Extreme Shepherding

We looked at the SeaLevel R9 single-board computer. Jim Slater pointed out that the Beagle Board has many of the same features, is smaller and quite inexpensive at $149. Jim will give a walk-through of the Beagle Board features next month.

When discussing about the September meeting, the consensus was that we would take the morning to attend the annual Halted Parking Lot Sale, then rendezvous for lunch at St. John's Grill on Lawrence Expressway.

Kevin put forth proposals to buy a set of wireless microphones to improve the recording of Forth Day and purchase a new battery for George's laptop.

Sam Falvo gave a brief introduction to his blog software written in ANSI Forth.

Jason pointed out the Green Arrays website is now online.

John Rible brought some example ColorForth code - counting the number of "1" bits.

13:15

Computing Logarithms and Exponentials - Eric Smith
The most common algorithms to compute approximations of transcendental functions (including exponentials and logarithms) use polynomial approximations. Such algorithms are well-suited to microprocessors with high-performance floating-point hardware, but are extremely slow otherwise. Since many embedded systems use processors without floating-point hardware, algorithms that do not require multiplication are useful.

I'd previously said that my talk on computing exponential and logarithmic functions without multiplication didn't have any direct relevance to Forth. I've now got fixed-point exp and log functions working on PACE FIG-Forth, so I'll be able to give a demo.

The supporting words for my exp and log words have some dependencies on word addressing, so they'd require some minor changes for a byte-addressed processor. FIG-Forth didn't provide a word for double precision logical shift right (or even single precision), so I've written a fairly inefficient but portable definition of that. More modern Forths may have such a word (e.g., RSHIFT in ANS Forth), but otherwise I'd recommend implementing it as a primitive.

I tried installing gforth on Fedora Linux, but it was compiled for 64-bit cells, and I'd rather get the exp and log functions working on a Forth with 32-bit cells since that's presumably much more commonplace. I tried to get the RPM of gforth for 32-bit Fedora to install, but it appeared to be corrupted. Maybe eventually I'll have an ANS Forth version of the exp and log functions, but it won't be ready for the August SVFIG meeting.

14:00

Break

14:20

Five Topics for Discussion - Jim Harrison

1. Why floating point isn't "friendly" - it isn't associative - and how to make it so. My opinion: it isn't worth the trouble.

   Kahan summation algorithm

2. While I'm waiting for GreenArrays (GA), hello XMOS. I will bring an XMOS board and chips to show and tell. I already have the development board and hopefully the chips will arrive by Friday.

3. Community building in Forth? This is an invitation for people to collaborate on porting Colorforth to XMOS, including documenting the whole process. This will serve as a general tutorial for porting Forth. Why collaborate? I haven't started and I have never done anything like this before. Why XMOS? It has a simple instruction set, is not very expensive, and the chips are available.

4. Quaternions in Forth. I will use quaternion multiplication to illustrate "good" (and bad?) Forth programming. The discussion will start with browsing a thread that I started on the Colorforth list and continue with comparing Colorforth's virtual machine versus other Forth's VMs and how that impacts the coding style.

5. Opening session of the Haypress Creek Fan Club. I'm sure other SVFIG members have been reading Chuck's blog. I have questions, perhaps others do as well, so perhaps we can discuss them together.

16:00

Adjourn

Other items:

Forth

Chuck Moore: Geek of the Week

Mark Washburn's Forth in Java Programming Webpage

MyForth Java Applet

Ben Simon's blog:

Dabbling with Forth

Two Languages for the Price of One - 10/25/2007

Software

TECO

Hardware

Books

Other