posted by [personal profile] hendrikboom at 11:00pm on 20/11/2008 under , ,
[Error: unknown template qotd]

I feel sorry for them.
posted by [personal profile] hendrikboom at 04:45pm on 29/10/2008
Here's a meme that's going around, and one I can get behind:

Copy this sentence into your livejournal if you're in a heterosexual marriage/relationship (or if you think you might be someday), and you don't want it "protected" by the bigots who think that gay marriage hurts it somehow.
posted by [personal profile] hendrikboom at 11:24pm on 20/10/2008
Of all the new Year's resolutions I've ever made, this year's has been the hardest to keep.

The things that have to be done a little at a time on a regular basis, like ten minutes a day -- Do them.

What kinds of things? Things like reviewing Japanese vocabulary. Ten minutes a day is enough to keep one learning. Spending five hours in a single block memorizing is not as effective as if those five hours are spread out ten minutes a day over a month.

I'm just failing completely on this one. I'll may have to reresolve it next year. I think it's important.

-- hendrik
posted by [personal profile] hendrikboom at 07:53pm on 19/09/2008
about a decade or so ago, someone in the Russian defense system decided that although their electronics were indicating a massive nuclear attack by the Americans, he reported to his superiors that there was nothing special going on. And so the world was spared.

Now this was a specific person on a specific day, and it is now celebrated as a day named after him.

Does anyone remember the actual date and who this guy is? I seem to remember it's at the end of September of the start of October.
posted by [personal profile] hendrikboom at 08:34pm on 25/08/2008
Let me quote Hermann Weyl:

The continuum of real numbers has retained its ancient prerogative in physics for the expression of physical measurements, but it can justly be maintained that the essence of the new Heisenberg-Schrödinger-Dirac quantum mechanics is to be found in the fact that there is associated with each physical system a set of quantities, constituting a non-commutative algebra in the technical mathematical sense, the elements of which are the physical quantities themselves. (1928)

Now as far as I can tell, this is the idea behind the noncommutative geometries the mathematicians among quantum-mechanical researchers have been excited about recently.

What puzzles me is that it seems to have taken seven or eight decades to get to the point that people are taking this seriously. Could it be that operators like -i hbar d/dx don't fit physicists' physical intuition, but real numbers too?

I've had an idea like that myself, too, about ten years ago, but have not the technical facility and background knowledge to develop it -- that the operator representing momentum that is the real physical quantity, and that the real-numbers eigenvalues that we "measure" are just momentary glimpses into a deeper reality.

Isn't the idea that real numbers are the results of measurements also an experimental falsehood? All the experimental results I've ever seen have been rational numbers. Why assume that the actual physical values we can only measure approximately are real? Why not assume they're members of a noncommutative algebra?
posted by [personal profile] hendrikboom at 11:20pm on 15/08/2008
To get something to grow in my front lawn so that the rain doesn't wash all the ground away, I planted clover a week or two ago. Now there are little green dots showing up in the bare patches. They're pretty. They're growing where the grass won't. But within some of the bare patches, the clover seeds have been washed to one side by the rain before they took root. I may have to reseed those spots and hope.

Last year, the rain here has occasionally been intense enough to wash some of the grass out of the ground. I'm hoping clover will reduce the impact of the drops a little better, and that it has better roots. We will see.
posted by [personal profile] hendrikboom at 02:00pm on 12/08/2008
I was asked, what am I doing with Janus?

It's the intermediate code generated by my Algol 68 compiler. There's a SNOBOL program that doesn't quite translate it into 360 assembler (I say "doesn't quite" because I'm sure there are still bugs to get out. The compiler only correctly ran a part of the Algol 68 test suite, and I would be very surprised if none of the deficiencies were in the Janus translator.)

So I have two approaches
-- translate Janus to machine code (probably a new translator). or
-- bypass Janus and translate the Algol 68 parse tree to machine code (what I was doing originally before compilation limits in the compiler I was using to compile Algol 68 H forced me to break things up and use and intermediate code.)
In combination with that choice, there's a choice of a local code generator -- there are a few now that I could use, I'm thinking of LLVM and C--, but there are probably more. There I have another choice -- to use one or the other, or neither and, say, generate assembler myself.
The questions are: which is likely to be more work, and which is likely to generate better code.
Since Janus can be compiled on a line-by-line basis with very little context (after all, it can be done by the STAGE2 macro processor), it is feasible to compile straight to assembler, the usual process for compilers on Unix. But debugging assembler is *hard*. (thought: it might not be so bad now that the world has usable debuggers). LLVM provides a fair amount of (optional) syntactic and semantic checking on the intermediate code, which I suspect will deal with most of the idiot-level bugs (and mst bugs are at that level.)
I could: parse JANUS and do pattern-matching on the parse tree to recognize stuff and generate code accordingly. I could do that parsing and matching easily enough in C, or in Scheme.

I perverted one thing in Janus -- the nesting of procedures. Janus doesn't do it. The code I generate does. The reason is that when I reach a procedure body in the normal tree-walk of code generation, the internal data structure in the compiler is just right for processing the nested procedure. To move the generated code out of the enclosing procedure would have been awkward, given the poor facilities available at the time for managing large text buffers. 400K memories were rare, and hard to get even if available; putting them in temporary files to be reprocessed later would have been awkward too, given OS/360's style of file access. But this is no problem now. Gigabyte memories are becoming the norm. What a difference a few decades has made! So it's perfectly feasible to implement text buffers in C, and let Algol W generate code into them. I've already got most of the API for that, except that it writes it all into a file instead. If I can manage to do something with the new varying-length strings Glyn has put into his Algol W compiler, it'll become even cleaner (the existing way I handle strings to be written to object code is to enclose the actual text in another inner layer of quotes to avoid the fixed-length-string restrictions of Algol W). If I use Janus as intermediate code, even if I choose to translate it into LLVM, I won't have a problem with interfacing Algol W with C++, which might or might not go smoothly.
Instead of assembler, I might be able to access a kind of low-level code generator I threw together a few years ago to generate code directly into memory for immediate execution. Leave that for later, if ever -- gdb doesn't understand this kind of code.
Does LLVM even have the kind of data structures Pascal and Janus use, with variants and such? If it does, are the ways of initializing them well-defined? Or is it just a matter of hoping future changes wont break what happens to have been implemented?
posted by [personal profile] hendrikboom at 01:58pm on 12/08/2008
10-ninute freewrite

Joseph was not an ordinary warthog. He was a smooth warthog. The other warthogs were snooty and complained about his smooth skin. ou are a smooth warthog, they woauls say, whereas I am a warty warthog. But Jseph was not this is really stupid, says my internal edotor. This is the kind af crap only a nanowrimoer could write. And it's not even Novembers. Go away, go away, dratted editor, I cry aloud. Your place it not here! Thiw is freewriting. Whatever drivel I produce is the right drivel. Nothing else will do. Begone, to the pit of profanity where you can bubble and blaspheme all you want. You will be recalled when needed. But the Editor would not leavel. He complained about toe L at the end of the worl "leave", and about the e on the to. I grabbed my head and pulled at my hair. This would go nowoere, the edotor was now complaining about the metarecursion in my story. Ant that's still more of ot! he cried with clee. How can you bring yourself to put down page after page of this crap! I said, it's only ten minutes, not pages and pages. But it would be if you were a decent typist. Then at least I could complain about style and characterisation, instead of all these stupid typos. Hey, I said, Typos are not the issue. THe ate , I repeat, not the issue. So why are you bothiering with this? Your time is to come, when I have the next draft of my novel written! Go off and complain to schoolchildren somewhere that they are -- oh I don't care. Jsut leave me alone.
I picked up my pen,. arising from the keyboard, and sprayed corrosive ink on his cut-and-paste sword. He parried, and I ended up just covering the tiger lilies in black spots. Aha! he exclaimed sarcastically, YNow you'r eon the right track! Those ink blots fit natirally with the spots on the lilies! Why can't you do somehting like that in prose. You're only a second-rate artist, but that's a lot better than you are a writer. How can you call yourself a writer when you can't even get a cleanly spelled forst fraft?
I was now completely distracted from what I had inteded to write -- I didn't even fell like I was freewriting, which had been the original ecercisel I was running around the back lawn, where I had written so peacefully only a week or so ago, trying to conquer the infernal editor. And, by the way trying not to do physical damage to the lawn while ai was at it.


end of freewrite. I can't even edit out the typos, because they'rre part of the plot!
posted by [personal profile] hendrikboom at 01:49pm on 12/08/2008
Nothing written yesterday, the lucky day chosen by the Chinese for the start of the Olympics. I have great trouble doing *anything* ten minutes a day on a regular basis, whether it's practicing katakana, exercising, doing a small increment of writing, taking a shower, brushing my teeth, doing exercises, clearing away junk in the basement, well, anything. Or it doesn't have to be ten minutes. It could be twenty, or forty minutes. So that includes things that are strictly for pleasure, such as watching episodes of a favorite TV show. Something always come up. Spending big blocks of time on something, that's easy. It just has to be something that snags my curiosity, and suddenly I'm spending a day or three researching it. But a lot of things aren't do-it-all-at-once things. They are skills, and unless the same skills show up in a variety of all-at-once researches, I don't get to get good at them. Mind you, there are subjects to which I return time and again, with gaps of weeks or months. But big blocks of intensity aren't the best way to learn some things. Better are many short efforts, ten or twenty minutes a day.
posted by [personal profile] hendrikboom at 01:47pm on 12/08/2008
My dentist has a slide show in his waiting room. Lots of travel photos, and a few of his family. But most of them are interesting. He seems to have done a good job of editing them. Some of them, he told me once, are there because they contain a visual pattern that resembles a 'Y', which it the first initial of his name. it could be a tree, a branching river, cracks in rocks, but there's a clear "Y".

There was more I wanted to write, when I was brushing my teeth this morning, but I've completely forgotten it.


6 7