Orbital Owl @Zuph@octodon.social

Hello World #introductions

I'm Brad, 31, he/him, in the US Midwest, and an engineer working on cool space hardware.

Stuff I like and am good at:
- Electronics
- Embedded Programming
- Cooking
- Hiking
- Tabletop Games

Stuff I like and am getting better at:
- Knitting
- Sewing
- RATS RATS RATS
- Web Programming

Stuff I want to do:
- Start a commune with all my friendos
- Fully Automated Luxury Gay Space Communism

Thinking about that time I sneezed 0603 passives all over the lab

See a sale for a 128gb microSD card: "Holy crap, what a great deal! I should stock up!"
See related items selling for $4 more.

Huh, memory is just that cheap now.

All I want to do is take trains and buses everywhere! That's it! That's all!

Gotta say, having just spent $900 on car repairs, that shit's not fun

A new member of one of the labs I'm working with is a Hitchensonian atheist, and insufferable. In an attempt to get him to quit arguing with me, I let him know that I considered myself an atheist too, and that was a mistake.

One choice quote: "Don't tell me as an atheist that you seriously believe in free will?"

airplane wifi paid for by work, shitposting from 40000 feet

I'm trying to figure out how I can help folks (including myself) to do better at documenting the hardware/firmware/software we make.

I think my ideal documentation system sits somewhere at the intersection of code autodoc system, wiki, and bug tracker.

We should be able to document and understand caveats, remove them when they get fixed (and maybe document the removal!), and put weirdness on the to do list automatically.

I'm not aware of anything that's close to that right now.

This is a really cool string formatting library for C++: https://fmt.dev/latest/index.html
Borrows the Python 3 string formatting rules, much safer, but not a lot more inefficient than printf.

Might mess around later and see if I can get it working efficiently on embedded with some fixed-allocation string templates.

When I feel overwhelmed by the rest of it, I tend to retreat into technology. It's something I'm good at, can hopefully help me later, and is intellectually distracting.

Wishing I knew better ways to keep that, but still make progress against the rest of it.

It's in the repository directory. And it contains the different ASF versions as tagged commits in that repo.

And (almost) every directory contains an asf.xml file with information about what files are needed how and when and why.

Dunno what I'm gonna do with this info. Probably waste a bunch of time. But, huh.

Obviously they have this data, because Atmel Studio can do it, through the Atmel Studio ASF extension. So, what's up?

Well, you see, the ASF in the extension is organized differently.

You can download the extension from Microchip's site. VSIX files are zips, so just open it with whatever. All the metadata is in extensions/[version number]/content.xml.cache

But where's the data?

They ship a GIT REPO with the extension.

Today I'm tearing into the Atmel ASF to try and liberate it from Atmel Studio for reasons that I still don't really understand myself. And sheesh, this is some weird/clever/weird corporate obfuscation.

The ASF they distribute as a zip is basically useless. It has all the source, sure. But all the example project and doxygen files are mixed right in, making it /basically/ impossible to include in your project. Plus, there's no dependency data! No data on what goes with what!

I think for me that means licensing things separately for "personal, non-commercial, and small business" use and "commercial" use.

We should also think seriously about licensing that excludes parties with malicious intent. I wrote this a while ago as a joke: https://github.com/Zuph/ACAB-License/blob/master/LICENSE.md/LICENSE.md

But maybe we should take it more seriously. What do I want my work to manifest in this world? It sure as hell isn't Palantir.

been thinking a lot about open source culture and tradition since reading this (https://blog.licensezero.com/2019/08/26/but-you-said.html) especially wrt to hardware.

Hardware is different in that the cost of replication is non-trivial, yet to be considered Open Source Hardware you have to let people make money from your work.

So: How to license hardware in a way that is fair to folks, but doesn't exploit the labor of the person that designed it?

This is also the first design I've done in kicad in YEARS. Definitely some learning pains-- I wish I could grab and move parts and traces like in diptrace, especially.

This is an adapter to help me take a Dell MD1200 that I got criminally cheap and hook it up to some quieter fans, so I can stand to run it in my house.

The quiet fans don't move as much air, but this thing will never see anything resembling worst-case conditions storing my filez.

Really, all this does is translate the RPM values coming from the slower, quiet fans into a range the server expects to see.