1337p337

I loved that language. I actually forked it, used it for a lot of stuff, bloated it (started by just trying to port to x86-64, ended up with a mini-FORTH with regexes, FFI to C, etc.). I still use it every day, though mostly for doing math in hex.

Looks like you are correct, yeah, and the raw logs show 7,410 of the 10,729 live instances have it. Lemme hack it in real quick; I can't put it onto the big pages and not sure how well adding another column to that table will work, but give me like five minutes.

Believe it or not, the resource use is mostly bandwidth for that page. Something like 2MB of HTML; the actual data is mostly cached.

That'd be cool, but it just uses publicly available APIs for this. Most of the data like that isn't reported through the public APIs across various types of software. Bad data is tossed out (e.g., the you-think-your-fake-numbers-are-impressive.well-this-instance-contains-all-living-humans.lubar.me lists over 7 billion accounts) but data that is (as far as I know) accurate is presented as-is.

Ha, indeed. It seems to be up for me, but that's why I put it on the demo subdomain and left off HTTPS: it's a demo, not quite ready for production.

(It's also getting flooded with vulnerability scanners at the moment, but that's not affecting the load, just the pipe.)

This applies if you're thinking of Twitter as a business or as a tool for mass-broadcast. If you think of something with a similar format but that fits a different space, that's the use for the Fediverse: it's more like USENET or BBSs or even 2008 Twitter than it is like 2022 Twitter.

The main advantage of tools like this are that you can use them to communicate. When I don't have the need to do that, I don't use them. I try to use tools that will remember this sort of thing on their own.

I keep notes on ideas in a personal wiki (AwkiAwki), and I open up the recent changes page to remember what I was taking notes about. For anything past the planning or brainstorming stages, I just use `ls -t|sed 10q` on my directory full of editor session dumps. I always keep a scratch/memo buffer open in acme so there will be some brief notes, an outline of the behavior I planned to implement, or (more often) I can gather what was going on from which files were open and where the cursor is in those files. For non-code projects, there will at least be some text, so there will be a session dump.

Forth isn't super practical in such an environment; gForth in particular is a bit "large". (Not that this is bad, but it's of the second school the article mentions.) If you wanted, you could cobble together a Forth implementation that played nicely with pipes. I did this, it works fine, but it's not the type of environment that matches Forth very well.

Forth's design is to give you the machine and let you compute with it rather than to act as part of a pipeline.

Yep, that's why I left it off the list.

I've actually written some LambdaNative. Unfortunately, we ended up going with Cordova, but you can guess which one was more fun to write.

Stock Scheme does not have much in the way of interfaces to the outside world (long story) but there are several Scheme variants where people are doing larger projects. Racket is sort of a Scheme variant, and it seems to be the most popular, but Chicken and Guile are also pretty popular (as far as Scheme variants go) for real-world use cases.

Not a false dichotomy at all. Ignoring that the size of the team is exactly two (and there's only so much you can keep in your head), when you have a metric, you optimize for it. It's not even a conscious process.

Well, PG presents them as different bets, one of which paid off. They were the same essential bet from Newton's perspective. For example, books on physics at the time referred to religion heavily, sometimes as a cornerstone of an argument. Even Newton's Philosophiae Naturalis Principia Mathematica says "Collocavit igitur Deus Planetas in diversas distantiis à Sole, ut quilibet pro gradu densitatis calore Solis majore vel minore fruatur."