Another fun fact: there are three commonly used ways to define week numbers (see `man strftime`, %U, %V, %W; %V is the one used in at least the Nordics). In some years they coincide so you might not notice that you picked the wrong one until next January.
Yet another fun fact: with %V week numbers, the date 2024-12-30 (December 30, 2024) was 2025-W01-1 (the Monday of week 1, 2025). Thus strftime needs two different ways to specify the year: %G denotes the year that goes with %V week numbers, %Y denotes the year that people usually think of when they ask "what year is it". Unfortunately %G comes before %Y on the strftime man page, so people who scan the page quickly can easily pick %G when they really want %Y. I've seen a few bugs caused by this.
I have also seen the corresponding bug in SQL, using IYYY instead of YYYY. This boggles the mind, but apparently when some people read "ISO 8601 week-numbering year", they only see "ISO 8601 ... year", think "yes, that's the date standard we use" and don't care about the "week-numbering" word in the middle.
I've seen Greek computer scientists write Omega as an underlined O. Much easier than trying to approximate the printed letter Ω but still unambiguous, unless you use underlining to denote something in your formulas.
This game (under the "Mafia" name) has been popular in Finnish math-contest circles since some time in the 1990s, probably as a Russian import. In large groups there can be many more roles, such as the axe-wielding lone killer, the police chief who gets reliable information from the game master, and the doctor who can rescue a victim if they guess correctly. Lots of fun.
Someone mentioned Blood on the Clocktower <https://bloodontheclocktower.com/> which has many more roles and a more complicated game that can take hours. The upside is that you aren't out of the game when you are eliminated.
I also learned about Mafia from participating in math competitions (in the US)! My teachers were Turkish immigrants and they introduced us to the game and we had a blast. it did suck for the person who died on night 1, though.
Quixotic[0] (my content obfuscator) includes a tarpit component, but for something like this, I think the main quixotic tool would be better - you run it against your content once, and it generates a pre-obfuscated version of it. It takes a lot less of your resources to serve than dynamically generating the tarpit links and content.
Likewise. My last name contains a non-ascii character. In ~2009 I started at a company whose admin conveniently set up an account for me on their Ubuntu server... on which no-one could then log in locally because the login manager crashed when trying to display the list of users. I logged in via ssh and changed my name to the nearest ASCII equivalent.
I always feel slightly worried on sites that demand that I give my full legal name (such as the US ESTA form), and then refuse to handle it because it includes "illegal" characters.
"Legal name" is a catch-all term that usually means "approved for use on government issued ID". Are there instances when that's not always the case and some forms of ID (not just, say, an ID card, but also in tax filings, for example) actually have different rules? Amazingly, sometimes yes. But usually that's what it means.
Legal system as in court of law? They tend to use more letters than I have in my actual passport (definitely more than fits into mrz) and depending on which court we talk about they also use different alphabets. They also assume certain structure in those nsmes, which differs from one court to another.
Yes, I had a pleasure do deal with two courts that use two different alphabets this year. They one of the two referenced the other. The name written in neither of two matches whats actually written in my passport. It isn't a complicated name by any reasonable metric.
Taxes are easier -- they just ids and names are display only kind of stuff, sourced from the base registry.
This has happened to me with passwords containing foreign characters. The system would accept it, but further logons would be impossible. Now I always strip diacritics to be safe.
A friend mentioned using control characters in passwords... like ^F and ^B, but not ^C because that's the interrupt character. Feels vaguely risky to me (does ^U empty the line? does ^W delete the last word? does your terminal emulator do some weird encoding like it does for cursor keys?) but if it works, why not?
I suspect I have run into a couple bugs because of password generators putting characters that some backend system cannot process in the password. Halfwish they just did DKWhhjwqjkwqjmHSJKHAIUHQwdmlsadkl instead.
PETSCII? On the Commodore 64 you could press the Commodore key and Shift together to change character sets between lowercase and the graphical characters.
But the Unix login thing might have been because of teletypes? https://www.columbia.edu/cu/computinghistory/teletype/ claims that ASR 33 used 8-bit ASCII but was uppercase only - not sure if the "8-bit" claim can be true.
On some Unix (and Linux) systems, you can still enter a kind of retro mode with "stty olcuc iuclc" (output lowercase to uppercase, input uppercase to lowercase) and turning on Caps Lock.
Yet another fun fact: with %V week numbers, the date 2024-12-30 (December 30, 2024) was 2025-W01-1 (the Monday of week 1, 2025). Thus strftime needs two different ways to specify the year: %G denotes the year that goes with %V week numbers, %Y denotes the year that people usually think of when they ask "what year is it". Unfortunately %G comes before %Y on the strftime man page, so people who scan the page quickly can easily pick %G when they really want %Y. I've seen a few bugs caused by this.
I have also seen the corresponding bug in SQL, using IYYY instead of YYYY. This boggles the mind, but apparently when some people read "ISO 8601 week-numbering year", they only see "ISO 8601 ... year", think "yes, that's the date standard we use" and don't care about the "week-numbering" word in the middle.