timorl @timorl@qoto.org

.hg so I'm banging some code out, right

#rust is complaining that a certain type isn't deserializable. I don't particularly care yet, because I'm trying to fix errors in other parts of the code, so I slap a `#[serde(skip)]` on there. /That/ doesn't work because the type isn't defaultable either. The obvious thing to reach for in this case is `unimplemented!()`, but that won't work; serde attributes only accept bare paths.

Since I'm trying to slap together at maximum speed the minimum stuff that will compile—and come back to it once the other things are done—I ask our brain for a function (not a macro!) that takes no arguments and always diverges. `_: fn() -> !`. Do we know of any such function that already exists?

Why, yes we do, in fact: `std::process::abort()`.

```
#[derive(Serialize, Deserialize)]
struct Foo {
#[serde(skip, default = "std::process::abort")]
bar: Bar,
}
```

:up
^Z
cargo check

my therapist: and what do we do when we feel like this?

me: identify the oppressive social systems responsible and dismantle the fuck out of them

my therapist: oh yeah that's way better than what i was gonna say

One meter is technically defined by the speed of light and a Cs-133 clock at 9.2 GHz, but it's not very practical. For real-world length measurement tasks you use light interferometry. But how do you obtain traceability from a microwave frequency to visible light?

Easy! Here how: Lock a hydrogen clock to the cesium clock first because for better short-term stability, then lock a 100 MHz crystal oscillator to the hydrogen clock to get started. Then, lock a 22.7 GHz Gunn oscillator to the 100 MHz crystal oscillator with a mixer and a PLL. Next, lock a 386.5 GHz Backward Wave Oscillator to the Gunn oscillator. Next, derive a 4251.76 GHz methanol laser from the BWO. Next, derive the 13C 16O2 laser at 29770.665 GHz from that with a mixer, a counter and a 8.99 GHz Gunn oscillator. Next, derive a 29477.165 GHz CO2 from that with a mixer, a PLL, and a 73.35 GHz Klystron oscillator. Next, derive a 28464.674 GHz laser, which derives a 28464.684 GHz then a 29477.165 GHz laser with a mixer, .... #electronics

agda (not to be confused with ada)

In a summary of the paper on Cambridge's security research blog, the researchers explain that even if you start with unbiased - that is, broadly representative - data, the *order* in which you present that data to a machine-learning model can induce bias:

https://www.lightbluetouchpaper.org/2021/04/23/data-ordering-attacks/

That's because ML models are susceptible to "initialization bias" - whatever data they see first has a profound impact on the overall weighting of subsequent data. Here's an example from the researchers' blog:

9/