Weeknotes: 20th February 2023

The week in review

I mostly worked on the MVP tickets:

I updated our auth server, paged back in how yubiHSM and Signatory worked, and installed the staging private key on the yubiHSM (after nuking old test state). I set Patrick up with access too, so there’s two of us with working access now. Patrick’s looking at deploying to Iona as part of our staging server set up first, and from that we’ll know if we’re good for production.
I wrote up our key management strategy, which is in this PR. It took longer than I’d expected as we as a group haven’t ever written up anything with the MVP architecture or use cases (at least that I could find), meaning I ended up having to write a description of our stack and how (I assume) it’ll be used. This is just waiting feedback from Patrick and Robin and hopefully can be merged.
I put together an initial "good/bad/ugly" list of Tezos development experiences, which Patrick and Robin have helped flesh out, as requested by Anil. This will go to the Tezos funding report, being worked on by Sadiq.

I think at this point I’ve moved all my 4C MVP tickets forward to where they’re either in review or being looked at by others.

I spent some time trying to help an ecologist with trying to get the radian R environment to respect $TMPDIR, but in the end used this as motivation to move our compute server to using tmpfs (/tmp as per user RAM disk) as per Anil’s suggestion a few weeks ago.
I did a little more learning OCaml, but didn’t make a huge amount of progress due to time.

Move forward the MVP work - annoyingly most those tickets still are in the “in progress state”, and I just want to nudge them to done so I can forget about them again.
Look at some issues Ali hit with her latest crank of the handle on the H3 tile work - a couple of species hit errors that we didn’t get before. That may be because she’s running a slightly older version of the code that’s not matched to the latest version of yirgacheffe changes I made, but it’s too early to say. It also seems to have been dragging on, so I’d like to see if we’ve had a performance regression somewhere.
I need to conclude the database evaluation work I started last week.

I’ve no had chance to dive on these, but came across these bits of related work to my ideas on Yirgacheffe 2/Ark:

Starlark: a simplified embeddable dialect of Python with Go/Rust backends
Meta Programming System: Jetbrains seem to have a way to describe DSLs that work with their IDEs, as a way of building programming systems for non-programmers.