We have a non-truncatable currency system. An ideal coin system will have the following characteristics - that it is "countable" for any number, and that eliminating the smallest coin will always leave the remaining coin system countable. "Countable" means that that cashier's algorithm of pulling the largest coin less than the amount needed and then repeating with the next largest coin will produce the optimum number of coins. For example, with our current system, if you need to make 42 cents, you do a quarter (leaving 17 cents), then a dime, (leaving 7 cents), then a nickel and then two pennies. Our current system is only truncatable for the penny. If you only consider coin amounts which are multiples of 5 (since other ones become impossible after getting rid of the penny), our system minus the penny is clearly countable.

The problem is, consider getting rid of the nickel. Now try to make 30 cents. You pull a quarter, leaving 5 cents, oh shit you made a mistake. Back up, you should have done three dimes.

Microsoft deemed it such an important project that they gave it an arbitrary trademark. For the most part Microsoft doesn't do arbitrary or fanciful trademarks (they do descriptive marks like "Visual Basic"), it's just part of their brand strategy. The name itself was come up with by brand consultants, who are basically one category of worker that has been almost completely supplanted by chat GPT and no I am not kidding. The company says what they want the brand name to evoke, a long list of potential names is brainstormed (or chat GPT created), culled, checked for trademark issues then presented to management.

Steve Jobs always called software "apps." There's video from the 1990s of him using that awful word when talking about Mac software.

DOS 3.3 had a maximum partition size of 32MB. If you had an older computer which had a small hard drive or no hard drive, ok but even the IBM AT released in 1984 had a 20 megabyte hard drive. DOS 4.01 was sorely needed when it came out, though if you already had a DOS 3.3 install there was no reason to upgrade to 4.01.

Though I am shocked that they open sourced 4.00 which was notoriously buggy. 4.01 was the release that quickly followed that everyone had. MS-DOS 5 was marketed as an upgrade for existing users, it included DOSKEY (which included command line history), EDIT, QBASIC and a lot of other stuff that made DOS less awful to use. Think MS-DOS 5.0 also introduced online help for the various commands.

There are way more sources of topical humor these days. Same reason late night TV has pretty much died. It used to be that if you were working at a factory in Iowa, if you wanted topical yuk-yuks you had David Letterman and theonion.com. These days, as soon as a news story hits, it's posted on Reddit and the comment section immediately produces every possible joke and then uses crowd-powered intelligence to upvote the best jokes. That's on top of Tiktok and Youtube creators, who often have staffs of their own.

Plus with respect to The Onion specifically, nobody reads actual newspapers anymore, or magazines, so a medium that gets its humor from aping the styles of print doesn't hit hard. I wanted to buy a copy of The New Yorker the other day and I had to check ... a large grocery store (which had the sign and the rack for Magazines, but which had been stocked with children's books), Target (which had some books but only a few magazines, mainly celebrity gossip), a bookstore (whose clerk informed me that their lease was up in 3 weeks and they had stopped bringing in magazines), and finally a SECOND bookstore. Frankly I'm amazed that any print publication can remain in business.

Early Postscript printers (the only printers that matter) had 13 "fonts" - which were three typefaces, Courier, Helvetica, Times Roman, each in roman, italic, bold, and bold-italic, plus Symbol. For decades Helvetica and Times were the standard fonts for newspapers and magazines, and Courier was similar to typewritter fonts. So when printers came to businesses, you got the same font choices. Courier was obviously not going to be used, except for marking up drafts or other specialty uses, and that left Helvetica and Times. Research showed that serifs helped ease of reading, so Times was the choice.

Now that does not mean that Times was the best choice for reading on-screen (the lower resolution caused the serifs to be a hindrance to legibility) and it doesn't even mean that of all serif fonts, Times was the most legible font, the most beautiful font or the best choice for business communications. It was invented for use in newspapers to be compact and save paper. But it was the standard.

Now here's a sidestory. The original font design was made by the Times (not the New York Times, rather the Times of London) ... in conjunction with Monotype, not Linotype. Their font was called Times New Roman. When Linotype made their version of the font, they simply called it Times (or Times Roman) and the font metrics were different. Times was the font that was licensed to Adobe and Apple for use in Postscript, so it was the standard. When Microsoft made Windows 3.1, they were too cheap to license real Times from Linotype, so they went to Monotype and had them modify the original Times New Roman font to be print-compatible with Times. So that documents created with one font would have the same line breaks and page breaks as the other. So Times New Roman is in a sense the original font, and in another sense a knock-off of Times. (The same is true of the other knock-off fonts included by Microsoft - Arial and Courier New.)

Microsoft, being cheap, until the mid-2000s never bothered to license any good fonts. The other fonts it started to ship with its products like Verdana and Comic Sans were even more hideous. As people need to use fonts available on all computers, the obvious choice is Times New Roman.

At a certain point, Microsoft realized enough was enough and gave us Calibri and Cambria which are not too horrible. Does that help?

The pointer with the reference count isn't the pointer to the object. It's the pointer in the "isa" field of the object which refers to the object's class. In the 32-bit ABI, objects don't carry reference counts in the objects themselves, it is stored in a separate hash table (which has to be locked prior to access). It's a legacy holdover from a time when memory was very scarce and putting a reference count into every object would be way too expensive. https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.mikeash.com%2Fpyblog...