Is it a lost cause?

I would be, since I am the upstream maintainer!

In 1985/6 in the UK we did 68K assembler in the first year of a BSc course. We didn't get down to timing diagrams. But we did do simple IO with stuff like timers and 7-seg LEDs. But then it was pretty much an introductory module.

Before that we had learned Modula-2 and ML.

I'm not sure if any places teach anything like that, that early any more.

In the same time frame, my PPOE taught 68000 assembler in the second year (before that, they did Modula-3 and other stuff).

I taught the operating systems modules from 1978 to 2015. At that time, I had them programming a basic bar code reader (getting interrupts on light/ dark changes, establishing a threshold for width, etc.).

Currently, they do AVR assembler in the second year (just too much else to cover in the first year, but they do get introductory OS and networking, which I taught until the end of 2015).

Very good to hear that this is going on... somewhere!

Over the years (not quite as many as yours) I've noticed that there is a divide where many programmers have no understanding of the hardware and vice-versa. It seems to me that only those who have (at least some) concept of that low-level 'bridge' really get the best out of both.

FORTRAN II has conditional jumps.

IIRC, it was a three-way branch: one line reference for < 0, another reference for == 0 and the third for > 0. This matches well for the machine code of early IBM machines, such as 1620.

Any self-respecting compiler should handle that easily enough, e.g.

For floating point tests (with the value in the FP accumulator) generate: BFP 0 LZERO [ zero FP value BFP 3 LLT [ less-than-zero code .... [ fall through to greater-than-zero code

For integer tests (with the value in the accumulator 7) generate: BZE 7 LZERO [ zero FP value BNG 7 LLT [ less-than-zero code .... [ fall through to greater-than-zero code

That's right. In practice, often two of the targets were identical, to construct all the conditional branches. This kind of usage shrinks to the BZ, BNZ, etc. There were no unsigned comparisons.

ISTR it was called the computed goto in FORTRAN 66

But 45 years since I last used it!

Wrong.

Pause for thought.... after all it was 45 years ago... Arithmetic IF

FORTRAN II did have the computed goto but the three way branch was the arithmetic IF.

Who said that the pi was aimed at "intending computer scientists"?

You conflate two separate discussion threads, one about what computer scientists should know, and the other about my interest in low level system programming with blinkenlights as the interface because it is at a lower level than VDUs and keyboards.

One of the main targets for the Raspberry Pi is high school children.

So? "high school children" != "intending computer scientists".

If the PI only teaches em how to set up apache and build a simple web site, its thoroughly useful.

Would this work for negative and positive zero?

/BAH

twos-complement scalar (binary integer) values don't support that concept.

If the Pi gets some of them started on a path of learning about something that fascinates them it will have done good. I see plenty of evidence that this has happened.