jdpage

Yeah, I don't see myself using this for Git (where it's very easy to see what you're adding, but not obvious and high-cost if you're missing something), but it's my standard approach for Docker images (where it's easy to tell that something is missing, and low-cost to fix it).

It interactively shows you each change that would be added and lets you decide whether it should be staged or not. Down to the hunk level, so you can partially stage a file if you so choose.

A few years ago I had a Windows box where I pointed XDG_CONFIG_HOME et al to the corresponding-ish Windows paths (AppData subdirectories, mostly) so that the many Linux-derived CLI programs I was using wouldn't clutter up my userdir, and it worked pretty well, though there were a few programs that didn't get the memo. Unfortunately I don't remember which ones they were. Most programs get it right, though!

I don't know what the official policy is, but glibc uses versioned symbols and certainly provides enough ABI backward-compatibility that the Python package ecosystem is able to define a "manylinux" target for prebuilt binaries (against an older version of glibc, natch) that continues to work even as glibc is updated.

The canonical Boolean values in FORTH are 0 and -1 (that is, all bits set). IIRC the point of that is to unify the bitwise and logical operators, though, not detect bitflips.

Also, at the machine code level, a Boolean controlling a branch or a while loop often doesn't ever make it out of the flags register, where it'll only be a single bit anyway because that's how the hardware works. Not really changeable in software.

I don't think it's that the lawyers are undifferentiated; some are better than others, in the same way some real estate agents are better than others (source: people complaining bitterly about theirs).

The problem is that there's currently no effective way to comparison shop. Needing a real estate agent or lawyer is a rare event for most people, so they don't have grounds to make a comparison. And you can't really tell how good they are until you've hired them.

Compare to, say, an auto mechanic. In the US, 90+% of households own a car, and after a few years of car ownership, unless you're a much luckier person than I am, you've probably been to a few mechanics and had some good and bad experiences. On top of that, most of your friends have done the same, so it's much easier to get enough data to figure out who to take your car to next. And, of course, the stakes are a bit lower: a lot would have to go wrong for you to end up in jail because you picked a bad mechanic, and cars are usually cheaper than houses.

Good to know! At previous jobs, that information wasn't available to me (and it didn't matter because the customer bases were small enough that every customer was top priority), so I assumed it was considered more sensitive than it perhaps is.

Tangential, but I wonder if the given example might be straying a step too far? Normally we want to keep sensitive data out of logs, but the example includes a user.lifetime_value_cents field. I'd want to have a chat with the rest of the business before sticking something like that in logs.

I have had many people use it when I try to either push for a private option ("please message me on Signal") or explain why I won't use a service.

Yes, the understanding is that trans people were always trans. It may have taken time for them to understand that and perhaps more to decide to adopt that identity publicly, but they're not "not trans" before that. Other queer identities are generally thought of the same way in queer communities: many people have early experiences (e.g. fixating on same-sex characters in fiction the way peers might opposite-sex characters) that they later realise were early expressions of their orientation.

Bear in mind, the terminal goal doesn't actually require unbiased numbers; the way most TTRPGs work is that you're trying to roll over or under a target number to get a weighted, unpredictable outcome. The idea is that while players (usually) want any given action to succeed, they some of their actions to fail in order to preserve narrative interest, while having their character be better at some things than others.

As such, while randomness is best, the given method is quite sufficient for having fun, and both players can agree that it's fair: they each have equal influence over the result.

No, it's trained into the model weights themselves.

This strikes me as a rather uncharitable view. I think it's okay for people to be proud of their work on a difficult project, and want to have their names on it.