IMvhO, the genius wasn't so much the implementation (which in hindsight follows the essence of the function), but the boldness that enabled him to even look for such a simple implementation. However from what little I've read about him in those days, that was the way I thought. Maybe it's a odd distinction.
EDIT: I do have to wonder about the survivor bias in the CNET story. If Apple had crashed and died like it almost did, we would never have see stories like this.
Yeah. I've read all of folklore.org and I bet there's an equal set of fascinating stories that were never written by all the companies that failed to do what Apple did. It's a shame.
- interview with Woz on the design: https://web.archive.org/web/20120212185115/http://apple2hist... (worth reading for the quote “We worked all night the day before we had to show it [the disk drive] at CES. At about six in the morning it was ready to demonstrate. Randy thought we ought to back it up” alone. Source control was quite different at the time)
He basically did super careful timing of the software to allow the main CPU to do the controlling. Because timing needs to be precise the software had to be perfectly timed down to the cycle. Woz's designs were very tightly integrated with both software and the electronics, and he would use every part of an IC to keep the count down.
I think he used some software routines on the 6502 CPU to do the work of sending data to the drive. This was very timing sensitive so be partitioned out some of the low level sequencing to the controller hardware while the rest was done by the very timing sensitive software routines. This technique is called bit-banging in general.
For some reason, the article reads like a high school writing assignment that required a minimum number of words. It reads like a puff piece. I can't figure out why.
Before anyone replies that I didn't understand the article, I programmed on the Apple //e back in mid 80s in 65C02 assembly.
35 days = 5 weeks = 25 work days = 1000 working hours (assuming a 40 hour workweek). that works out to $13/hour for a programming contract. even with inflation accounted for ($49 in today's dollars), that's low. am i missing something? maybe they only worked half the time (which brings the effective rate to $98)?
Couldn't find much on salaries for 1970s, but here's a resolution to hire a junior programmer for ~$10.6k/year in 1979. Seems like the $13k referenced was not a huge amount, but about an annual salary for a single dev.
