What are you doing this week?
11 points by caius
11 points by caius
What are you doing this week? Feel free to share!
Keep in mind it’s OK to do nothing at all, too.
Getting rid of CloudFlare.
Week 3 of $newJob. Refreshing my web dev skills and determined to ship something before everyone's holiday zone next week
Related: wsl is okay but windows is still awful
Where does WSL fall short for you? I left Windows for Mac because I prefer the Mac UX. For my use, WSL gave me the unixy things I wanted, so I'm curious about your experience.
Working on new ways to promote amber-lang.com language (Bash Transpiler), we are working on various stuff for the next major but we need more Rust contributors.
Heading back to the UK and rediscovering what a jumper/coat is 🥶
Company festive party this week at least, that'll help with adjusting to normal life again. Have a charging system fault on one car to chase down (hopefully a fuse, given I've changed BMS sensor & battery already).
Mostly catching up on normal life after being away, seeing what parcels have arrived for me whilst I've been away (and subsequently forgotten what I've ordered.) Got a bunch of Shelly matter relays to wire into ceiling lights to make more of the house "smart" controlled, and had a new boiler fitted whilst I've been away so need to double-check the Tado system is happy with that and all the radiators are bled/working as expected, etc. Possibly tune the heating schedule now there's a nice efficient new (slightly over-specced) boiler providing the heat.
I am currently prepping for a sanctum [0] 1.0.0 release!
After a year of dogfooding and getting most of my hacker friends to replace their tailscale or zerotier setups with this i am confident enough its ready for a 1.0.0.
I've also been working on tier6 [1] (think zerotier but over sanctum) and getting the Conclave [2] website up and running to host all official releases of projects related to sanctum (like tier6, or the library that implements the protocol etc).
Trying to motivate myself to dig into DS&A again. I set a personal goal of feeling "ready" to interview in Go by April 2026. I have been feeling particularly demotivated for the past 18 months or so (since I started my current job) which has made it difficult to summon the energy and discipline to see this through, but I am going to try putting some time into it this week.
Current focus is DFS / BFS. I am reading the relevant chapters in my DS&A books, implementing the algs, and then trying LeetCode problems.
Packing all my stuff to move! Then, unpack everything this weekend. Then I hope to do some writing this christmas, both work-related (research proposal) and personal-technical stuff.
Creating Daino Qt - a collection of components that makes Qt apps feel and look native on both Desktops and mobiles (each with its own set of challenges).
Developing Qt apps with C++ and QML is a blast - the fast performance of C++ and ease of use of writing UI in QML. But there is so much left to be desired with the built-in Qt Quick components - mobile issues like non native text handling, non native swipe-able stack view and much more. I’m aiming to bridge that gap.
Troubleshooting a degenerate decoding problem in my BitNet b1.58 LLM implementation. BitNet is an LLM optimized for running on modest hardware. I want foundational AI available to anyone, not just the ultra-wealthy and government affiliated. I intend to release the model in its pre-trained, fine tunes, step-by-step training states - everything - under an Apache license.
What dataset do you intend to use?
That's the question, isn't it! The model will be 1B and 7B to begin with. I'm considering https://github.com/togethercomputer/RedPajama-Data but need to look carefully at what's in it for copyright claims.
Writing my own HTML DSL for to_stringing in OCaml since TyXML has failed me for not implementing the spec correctly / kept it up to date—nor is maintainership on the ball for trying to merge requests (not that folks need to dedicate all their free time to their open source projects); I’m tired of the spaghetti of patches I have to keep it working that never apply because of diffs in the changelog. I don’t really think this sort of type-level safety is very useful as a living spec is not permanent & the type errors are insane to try & parse (something, something “…they didn't stop to think if they should”). Cow also failed to keep up to date & locked down certain elements—such as <script> not allowing async or defer. I could try to extend/patch it, but I throw up my hands & try to do it myself. If it fails, going to Jingoo (Jinga) templates & not looking back.
I should probably try to find another remote contract too :|
Reviewing some PRs for the MapLibre Native map toolkit.
This is a contribution from Meta and adds plugin system to MapLibre Native. The plugins need to be compiled into the library. That may make it harder to use for people that don't have in house build engineers (which is the majority of our users, even multi billion dollar companies prefer to use the builds that we put out). Also in this plugin system plugins are a Map Observer, which makes it awkward to register things that should really be global. We unfortunately have some singletons that would need to refactored first if we want all plugins to be per map, for example custom layers. Another question is: do we really want to expose the entire C++ API?
This is also a PR I need to review. Will probably help out with the render tests for iOS and Android. It ports some hillshading algorithms that were implemented for MapLibre GL JS before. See the example screenshots.
Preparing for a v0.12.0 release of https://rcl-lang.org/, which polishes a few minor rough edges that I ran into while using it for AoC. That's not what it's intended for, but it works surprisingly well.
Looks neat! I love Nix and really dislike the Nix language, so it's good to see that RCL takes some of the issues faced there into consideration. Slightly curious about the choice of the "make X Y again" phrasing on the front page, what led you to that?
Thanks!
The ‘boring’ is similar to the ‘boring’ in BoringSSL. Building applications and systems is what should be exciting, configuring them should be the boring part. It’s not the place to be clever, and it’s not the place where you want to be fighting incidental complexity like escaping or indentation issues. The tools should get out of the way, rather than creating new classes of problems. I have the impression that configuration was simpler in the past, but nowadays it’s somehow normal to use multiple layers of string templating to render yaml eDSLs that configure a service mesh of microservices … I hope that RCL can make this part boring again.
There is no relation to a certain slogan used by a certain political figure, if it reminds people of that then maybe I should replace it …
I really like this philosophy about configuration! Especially the point about incidental "excitement" like spacing/escaping. In my experience a lot of the complexity of configuration languages is historical, the result of layering more and more systems atop one another, so maybe a clean break is exactly what we need.
I will say, it does immediately make me think of the fashy cheeto. But I'm glad the underlying philosophy is so well thought through!
Working on a programming language that I’ll use to make a game, all in 7 days for the Langjam Gamejam.