Return the carriage, feed the line by Aaron Taylor

Editor's Notes

• #3: As with all good stories. This one doesn’t start with someone you know. In fact, it starts with someone you don’t
• #4: This guy. Introducing Gary Kildall. Author of the seminal operating system that we all know and love, CP/M. Remember it? Nope? Neither had I until I wrote this talk. It’s late 1973 and Gary here has a decision to make. It’s a decision that would go on to have massive ramifications for computer science and all those who practice it. It’s a decision that we still feel the effects of today, one that I think has cost the world billions in bugfixes, lost time, frustration and general hair-pulling. You see, Gary here is working on the future. But he needs to support the present and, more importantly, the past. Namely, this:
• #5: Behold! The mighty Teletype Model 33 This beauty in beige was one of the first user interfaces; an operator would type commands which were sent to the mainframe, and the responses would be typed out, automagically onto some paper. This baby was capable of printing a staggering 100 words per minute … that’s almost 85 baud
• #6: (Next slide)
• #7: Now there were a couple of interesting keys on the keyboard here, namely these two:
• #8: One of those is the carriage return key, and another is the line feed key. Sounding familiar yet? Now, this is 1973, remember and even those two terms are archaic. They’re direct imports from back when typewriters were the height of information technology, and for the curious, here’s a handy diagram of what those things actually do:
• #9: Move the lever up to move the paper up a line. Move it across to move the carriage, and the sheet of paper back to the start of the line. But why not just do both? Well, sometimes you want to overtype certain words to make things bold, or do strikethroughs and things like that. But either way, it definitely made sense to have control of both, separately.
• #10: But Gary’s OS, CP/M — which stands for Control Program for Microcomputers — was being designed for the terminal. Those monochromatic CRT displays that were popular with the computing crowd of the day. The problem Gary faced was this: what command should we send when we want a new line. On an electronic display he could use anything he liked. He knew the Unix guys were handling this with a single line feed command. But this wouldn’t work with the Teleprinters. It’d just scramble the output. So he did the only thing that made sense. Two commands would be sent. One to return the cursor or the print head to the start of the row. And then one to move it down a line.
• #11: The commands themselves were: Carriage return, line feed. Or CRLF Oh, and if you’re interested: CP/M looked like this:
• #12: And got bought by this guy a few years later
• #13: Who went on to build this
• #14: Which would ultimately lead to developers worldwide having to put up with things like this
• #16: My next story begins some years earlier, in the 19th century and starts with a conversation about what to print money on. I don’t have the details of that conversation, but what I do know is that once the decision was made, the result would have such deep ramifications as to change how each and every one of us thinks about programming.
• #17: This is an 1890 United States Federal Reserve Note. A Ten Dollar Greenback. It is 188.9 x 79.8mm; roughly 20% larger than the US bank notes in circulation today. And when it’s being transported around, it fits in a custom-designed, secure box.
• #18: This is Herman Hollerith. He had the misfortune of working on the 1880 United States Census. A laborious, error-prone task that he thought could be done much better. He would spend the next ten years working on an automated solution for the collection and processing of census data, and after almost a decade of work, and a patent, Herman Hollerith would change the world for good as he birthed this:
• #19: The Tabulating machine. It was a marvel of its time. Punched cards were read at the station on the left, which opened one of the lids on the sorting box on the right depending on the values of the card.
• #20: Herman was a true engineer. Only solving the problems that needed solving. He designed the cards to be compatible with the existing treasury boxes that were used for transporting money, and so he gave them the exact same dimensions as the bank notes of the time. It was a phenomenal success. Hollerith would go on to found The Tabulating Machine Company in 1896 and through a series of mergers they would become part of a consortium of companies known as the Computing-Tabulating Recording Company who would later, in 1924, rename themselves again. Although they didn’t get any more inventive in their naming of companies, this time it stuck. They called themselves International Business Machines. IBM. The card in use is a diagram of the same one used during the 1890 census. It had a 24 column, 12 row lattice. The punched holes were originally circular, and for those of you who remember your geometry, circles don’t really tessellate very well. But the clever people at IBM had a plan for this.
• #21: Square holes! Behold, the IBM punch card. What you’re looking at here is 960 bits of raw information, stored in 12 rows of 80 columns. They’d later up the rows to 25, giving a whopping 2,000 bits of capacity. This format went on to become the industry standard for computer programming, with cards like these being used to program in languages such as Fortran. And featuring heavily in the space program. So ubiquitous were these 80 x 25 cards, that when CRT screens started becoming commonplace as replacements for teletypes, and once they’d got over the nearly insurmountable problem of displaying up to 2,000 characters on the screen at once, the format for computer terminals became 80 x 25
• #22: And then storage improved and programming languages changed. You didn’t have to carry around bundles of cards with you any more. You could program straight into the terminal itself. And then screen resolutions increased. And GUI’s improved. But no one questioned the 80 column limit. But everyone kept on enforcing it. And so it stayed. So every time you see this in your editor
• #23: Or any time your linter tells you that you’re lines are too long. Just take a deep breath, and thank Herman Hollerith and his fantastic tabulating machine
• #24: My next story begins some years earlier, in the 19th century and starts with a conversation about what to print money on. I don’t have the details of that conversation, but what I do know is that once the decision was made, the result would have such deep ramifications as to change how each and every one of us thinks about programming.
• #26: It was originally a dumping ground for config files and things that no one could find a better place for
• #27: It was originally a dumping ground for config files and things that no one could find a better place for
• #28: It was originally a dumping ground for config files and things that no one could find a better place for
• #29: No one liked either of those tags, but both MS and NS had considerable weight in the fledgling W3C. A deal was struck to remove them both in February of that year, however Microsoft reintroduced the Marquee tag in 1996 with the launch of the new Internet Explorer