James Dean Palmer

I have an incredible habit of discovering great TV and radio shows just before they say farewell. I began listening to Penn Jillette’s Penn Radio only a month before the series would end. The radio show which adopted the seemingly rare libertarian-atheist point of view, played the issues of the day against arguments of reason intelligently and humorously. With some effort, you can download all the old episode that aired from http://www.pennfans.net/ for as long as that lasts. And although most talk shows don’t play well after their original airing because of their topical nature, Penn Radio really does since it is littered with fabulous interviews and guest hosts.

Similarly, I began listening to a Canadian radio show And Sometimes Y which has also come to an end. The show humorously considers language, words and literature with a geeky pop-culture aesthetic. Some of the old episodes are available in mp3 and RealPlayer format on their website but then there’s also this recent episode of CBC’s And the Winner is.. podcast which features one of their best episodes, The History Of English In 28 Minutes.

I really don’t enjoy reading books on a computer screen. At times I have been forced to read books online. At other times I have tried to force myself to embrace the idea. Recently a suggestion was forwarded to the faculty that we should really consider etextbooks as a more economical solution for students as tuition prices soar and the economy sags. Students can and should do what they like but I don’t plan on encouraging the idea.

A survey documented by Nicole Allen of 504 Oregon and Illinois college students found the following:

1. eTextbooks were no cheaper than traditional textbooks given most students buy used and/or sell back their textbooks
2. 3/4 of students said they would rather use a real textbook if it cost the same as an etextbook
3. only 1/3 of students said they were comfortable using etextbooks
4. access to many (3/4 of those surveyed) etexts expired after 180 days
5. many etexts disallow printing

The report suggests open textbooks as an alternative but this really only addresses the last two points. Other issues not raised in the survey include:

1. accessibility and portability may be limited as access may be restricted to where there is a computer and an internet connection
2. markup capability is more important for etextbooks than many other ebooks as students (and profs!) use highlighting, margin notes and stickies as learning and exploration tools yet is noticeably absent or noticeably miserable.
3. many very high quality texts are simply unavailable as etextbooks

Many students, computer science majors especially, spend enough time reading text on computer screens. There are social dynamics to using real artifacts like books and getting together with people in the flesh and talking and working over a book. I’m prepared to admit other recent innovations like the Amazon Kindle may make many of these points moot but many eTextbooks aren’t even compatible with the Kindle because of DRM incompatibility or restrictions. eTextbooks may be the future but I just can’t recommend them now.

Ginger supports type annotations as part of it’s design by contract support. These contracts may be enforced or checked at compile time if enough type information is available, otherwise they are checked at runtime. Type annotations can become part of the signature for the function (thus allowing overloading) for required arguments (but not optional and residual arguments). Consider this example:

1
2
3
4
5

define add (a b)
  require (is a <number>)
  require (is b <number>)
  ensure (is result <number>)
  + a b

We can of course make a type requirement for an optional third parameter but since it is optional it inherently makes the signature ambiguous and thus doesn’t count toward the function signature used in overloading functions. It still remains part of the contract though and would be enforced and in optimized production code allows the compiler to make useful assumptions.

1
2
3
4
5
6

define add (a b optional: (c 100))
  require (is a <number>)
  require (is b <number>)
  require (is c <number>)
  ensure (is result <number>)
  + a b c

I have finalized how function signatures should work in Ginger. My Ginger to Scheme compiler basically mirrored Scheme’s argument conventions but I’ve put a little more thought into it for the Ginger to C compiler. First an example where arguments are not named:

1
2
3
4

define foo (a b)
  noop
 
foo "A argument" "B argument"

The same function, foo, implicitly defines named arguments so the following also works:

1

foo b: "B argument" a: "A argument"

Now consider the definition of optional arguments:

1
2
3
4
5

define foo (a optional: (b "B argument"))
  noop
 
foo "A argument"
foo "A argument" b: "B argument 2"

And finally unnamed arguments are captured as a residual (called rest in many other languages):

1
2
3
4

define foo (a optional: (b "B argument") residual: c)
  noop
 
foo a: "A argument" b: "B argument 2" "and this" "and that" "and even" 1000

My truck has been unusable for a couple of months now. The master clutch cylinder and the slave cylinder were in such bad repair the clutch goes all the way to the floor and it had become impossible to switch gears. I was planning to fix it myself but when I realized I would have to drop the transmission to replace the two cylinders, I thought better of the idea. I had it towed into the shop on Friday.

We have been missing it for a while. Last weekend we were planning on going out to Grand Falls but without the truck we had to take my Saturn. We made it about 40 yards in the Saturn on “Indian Road 70″ before thinking better of the idea. The last leg of the journey to Grand Falls is pretty rough. It’s not so much that you need four wheel drive as you need a vehicle with tough suspension, a large wheel radius and the ability to take a few rocks to the head.

Something like my own F350 diesel behemoth.. I got a call from the mechanic this afternoon that my truck was ready. In turn, I called Yaya and asked her if she was up for trying out Grand Falls one more time. We headed out after work and made it just before sun set. The bumps and rocks near the end of the trip were the worst and reminded Yaya of the “Pink Jeep Tours” you can take near Sedona. This was more of a Black Truck Tour.

Yaya has some great shots on her blog but since it’s private I will go ahead and repeat them here:

These Doctor Who felt dolls by AGirlCalledRic are outrageously cute! I found her work on Etsy which is the ebay of all things homemade.

Shuttle has updated their web page with what may very well be my next computer. The KPC is small, lean and green. It sucks 33W of power when idle and 55W when going full bore (that’s better than my MacBook). AND it has customizable faceplates.

I have committed some preliminary work towards an “eval” function. The function itself is more or less complete except that it runs into an architectural flaw in ginger. Currently, ginger takes advantage of the fact that Scheme’s representation of functions and variable names work very similarly to her own and she purposefully maps between the two which would be fine except this mapping happens at compile time. Eval being an inherently run-time kind of thing needs this same mapping to happen at run time. Which I could do.. But it would be wasted code. Ginger needs her own way of looking up functions, namespaces and scope independent of the scheme runtime. In part, this is necessary to support ad-hoc polymorphism which will be much more natural and automatic in Ginger than it is in Scheme.

Why am I working on eval? With eval implemented, macros are easy. With macros implemented, everything else is easy.

For the past few years Yaya and I have had two TiVos. One original Series 2 with a 120G drive and one Humax with an 80G drive and an integrated DVD player and recorder. We really like the DVD player to the point that it would seem almost a loss to upgrade to a series 3 and lose the ability to play DVDs! We bought an HD TV only when our last TV’s image had gotten so discolored and blurry that each time we watched TV we questioned if our vision was going. But I am too thrifty to pay a premium for cable service or satellite service to get HD and then pay even more to get a series 3 TiVo to record it.. No, HD will have to take me kicking and screaming. But I digress..

The two TiVos were fine but it did cost extra money. The extra $6 for the extra TiVo service and an extra $5 for satellite service and an extra 40-50 watts of electricity being consumed 24/7. The obvious TiVo to retire was the original Series 2 but my 80G Humax is just a bit underpowered in terms of space. I could probably justify buying a hard drive and claim a 1 year payback on my investment but I already had a spare 250G drive gathering dust in a closet so it’s just a matter of coming up with the time to do it. And I have actually been meaning to do this for over a year but almost everything else was higher priority. Sometimes you have to say screw priority, lets just do it.

I started with this HOWTO whose shear length made me question the practicality and wisdom of an upgrade but once you trim out all the information that doesn’t apply to your TiVo, ignore all the warnings and information written for people who have never opened a computer and then do the upgrade process via Ubuntu instead of the boot disk (which I never got working) you really have a pretty short list of things that have to be done:

1. Open case and remove drive.
2. Put drive on your linux box.
3. Run the mfstools command to backup the drive.
4. Put new drive on your linux box.
5. Run the mfstools command to restore and expand to the new drive.
6. Reinstall drive in TiVo and close her up.

Easy peasy.

I have hastily put together a Ginger development site at Ging3r.org. Almost the only thing there is the Ginger repository which I am hosting with Mercurial.

My first official commit to the repository adds supports for includes. You can now say

1

include "utility-functions.gin"

and it does what you probably expect it would do. There are several ways this could be implemented. We could (a) “include” the file during the parse step, (b) include it immediately after the parse step or (c) make it an actual function that glues the code into the runtime environment.

C-style #includes employ technique (a) as the c preprocessor effectively “pastes” in the include before the file is parsed. PHP style includes work very similar to (c) and allow conditional file loading. For now, Ginger uses approach (b). It looks for includes in the parse tree and then parses each included file it discovers and glues the subsequent parse trees into the parent parse tree.