Post History

`/dev/random` is a stream of every possible value. You're supposed to filter it to take the ones you want. This is efficient, although obviously if the values you want are such that only 1% of what comes out of `/dev/random` is acceptable, then it will take 100x longer to generate them. But luckily `/dev/random` is very fast.

First, the foreword: `cat /dev/random` (yes, useless cat, whatever) will produce a forever stream of random data. This isn't very useful, usually you want a limited stream, you do it with `head`:
```
$ cat /dev/random | head -c 20
NL(c%t6ޖ[\{I⏎
```

This is taking 20 random chars. `head` is nice in that, even though the input is infinite, `head` will shut it off once it has what it needs (well, your shell will, whatever). You'll note it's actually 12 chars - head is getting 20 *bytes*, then they get rendered by my shell, some end up displayed as multi-byte Unicode chars. This unicode business will stop mattering in a minute. The newline is just added, it's not coming from `/dev/random`.

We now need to throw out characters we don't want. This must be some incantation that will produce only "good" chars, so that `head` can collect 20 of them and exit. The OG way is `tr`:
```
$ cat /dev/random | tr -cd A-Za-z0-9 | head -c 20
yBJ1Uh7yatwdZmgxVK8C⏎
```

`-c` is `--complement` aka invert. `-d` is `--delete`. So we have "delete everything that is *not* `A-Za-z0-9`".