Having a week of vacation in a nice lake hotel in the countryside. Starting in a new job at Grafbase next week. Dedicating my time to WASM programming for the next few years, still using Rust as my main language.
I’ve been on the drawing board, figuring out my language’s generic syntax. Specifically how to deal with multiple sorts of parameters (values, ADTs and HKTs) in a concise way.
Other than that I’m looking to get a better understanding of unique types.
Trying to learn my way around the Poisson process for this queueing theory reading group I’m in. My longtime weakness in understanding probability is holding me back.
I am nearing completion of the lexer for a RIIR project for the TLA+ tools. The language has A LOT of tokens so it has been a slog. I tell myself that actually constructing the parse tree for all those tokens will be a lot more fun.
I’m in the process of defining a toy-programming language that enforces - by design - cryptographic constant time (CCT).
I need to evaluate which way of ensuring this fits best in my case: I’m reasoning about a compilation pipeline and in one of the later stages I want to bring CCT to the table. Unfortunately, it looks like we need to argue about CCT even in earlier stages of the compilation pipeline. It’s more or less obvious: We need to know what values are “secret” and what are “public”. We cannot get this out of thin air at some later stage during compilation (although, given this is theory work, we may just assume these annotations “exists”)
It’d be nice to have a way to just glue more or less meaningless attributes to parts of the source programming language, which are carried through throughout the pipeline. But, this cannot really happen “meaninglessly”, because predecessor compilation passes may optimize stuff away (e.g. simple constant prop) or add identifiers (e.g. anf conv). So, we surely need a cross-language relation which associates late-stage attributes with early-stage attributes, to make sure they “stay the same”…
It’s a bit annoying, because we cannot do this in modular fashion. Maybe something pops up while looking deeper into it
Can we replace these with something retrospective? If I’m going anywhere in a week (for example, I was at Cyber UK last week), I don’t really want to post something publicly readable on the Internet saying ‘I won’t be in my house this week’, so I often don’t post in them when I’m doing something interesting.
$work: Getting a bunch of smaller tasks knocked off the todo list at work, then into upgrading Ruby in all the repos (most use it as a helper script language so not so difficult to update those.) Also getting severely tempted to restructure our terraform monorepo, to further break down our workspaces beyond what we currently have.
!$work: Attempting to build a results/presentation view into the scavenger hunt app, and possibly a photo gallery everyone can access once the game is finished. Lets all participants see what each other submitted for the results, and we want to present some snapshots at all hands once the game is finished rather than just “$x won”.
Also looking at upgrades for the BMW Z4, mostly around refreshing suspension. It’s just ticked over 113k and I nearly replaced it with a MX-5 NB but decided not to in the end. Given I’m keeping the Z4, I should make it nicer.
Working on a paper for MICRO. The problem with working on cross-layer research projects is that no one bothers to make sure deadlines don’t line up for different fields, so this comes just a week after the SOSP deadline.
More messaging work for https://rpgportrait.app, this time adding a series of “quick tips” to display on the loading page while users wait for their portrait to be generated. Mostly suggestions for how to get better portraits, but with the occasional reminder that there are more features available to subscribers. :)
Last week of the first month of new $dayjob after more than half year burnout recovery. I am still way more tired than I’d like, but people are super nice so far and that sure helps.
For $fun, I am trying to subsample periodic signal in a pseudo differential fashion with interleaving that induces a phase offset. I really need to simulate the approach with scipy before trying things out in C.
Hacking on my side project and developing a small open source HTML generation library alongside it. Using it in my backend. No weird templating languages needed!
Trying to make a first release for my pet project Spatial Shell. The code is ready, the man pages are ready, I think i just need to come up with a nice blogpost giving a tour of the project in a visual way to call it a day. What’s nice is that it is a project I have been using daily for almost six months now, which is always super neat (having a project driven by a very concrete need, rather than a fun ideas as some of my previous hobbies were).
Having a week of vacation in a nice lake hotel in the countryside. Starting in a new job at Grafbase next week. Dedicating my time to WASM programming for the next few years, still using Rust as my main language.
Finishing up the MVP for my AI startup before I start looking for funding.
Helping a friend do some data analysis to win a lawsuit against a stalker.
Launching a preview of my programming language at https://scrapscript.org and submitting my talk to Strange Loop.
Planning art installations for my tiny tech conference: https://outland.sh
Wonderful todo list!! Excited to follow along!!
Preparing a new release for https://systeroid.cli.rs
Wow, I would’ve loved using this when I was doing more sysadmin stuff a few years ago!
I’ve been on the drawing board, figuring out my language’s generic syntax. Specifically how to deal with multiple sorts of parameters (values, ADTs and HKTs) in a concise way.
Other than that I’m looking to get a better understanding of unique types.
Trying to learn my way around the Poisson process for this queueing theory reading group I’m in. My longtime weakness in understanding probability is holding me back.
I am nearing completion of the lexer for a RIIR project for the TLA+ tools. The language has A LOT of tokens so it has been a slog. I tell myself that actually constructing the parse tree for all those tokens will be a lot more fun.
I’m in the process of defining a toy-programming language that enforces - by design - cryptographic constant time (CCT). I need to evaluate which way of ensuring this fits best in my case: I’m reasoning about a compilation pipeline and in one of the later stages I want to bring CCT to the table. Unfortunately, it looks like we need to argue about CCT even in earlier stages of the compilation pipeline. It’s more or less obvious: We need to know what values are “secret” and what are “public”. We cannot get this out of thin air at some later stage during compilation (although, given this is theory work, we may just assume these annotations “exists”) It’d be nice to have a way to just glue more or less meaningless attributes to parts of the source programming language, which are carried through throughout the pipeline. But, this cannot really happen “meaninglessly”, because predecessor compilation passes may optimize stuff away (e.g. simple constant prop) or add identifiers (e.g. anf conv). So, we surely need a cross-language relation which associates late-stage attributes with early-stage attributes, to make sure they “stay the same”… It’s a bit annoying, because we cannot do this in modular fashion. Maybe something pops up while looking deeper into it
Can we replace these with something retrospective? If I’m going anywhere in a week (for example, I was at Cyber UK last week), I don’t really want to post something publicly readable on the Internet saying ‘I won’t be in my house this week’, so I often don’t post in them when I’m doing something interesting.
$work: Getting a bunch of smaller tasks knocked off the todo list at work, then into upgrading Ruby in all the repos (most use it as a helper script language so not so difficult to update those.) Also getting severely tempted to restructure our terraform monorepo, to further break down our workspaces beyond what we currently have.!$work: Attempting to build a results/presentation view into the scavenger hunt app, and possibly a photo gallery everyone can access once the game is finished. Lets all participants see what each other submitted for the results, and we want to present some snapshots at all hands once the game is finished rather than just “$x won”.Also looking at upgrades for the BMW Z4, mostly around refreshing suspension. It’s just ticked over 113k and I nearly replaced it with a MX-5 NB but decided not to in the end. Given I’m keeping the Z4, I should make it nicer.
Working on a paper for MICRO. The problem with working on cross-layer research projects is that no one bothers to make sure deadlines don’t line up for different fields, so this comes just a week after the SOSP deadline.
More messaging work for https://rpgportrait.app, this time adding a series of “quick tips” to display on the loading page while users wait for their portrait to be generated. Mostly suggestions for how to get better portraits, but with the occasional reminder that there are more features available to subscribers. :)
Beginning to write the next chapter of https://proseforprogrammers.dev on writing structures.
I may even get back around to writing my steampunk fantasy novel.
Adapting exercises from my grep ebook as an interactive TUI app.
Currently reading “The Steerswoman” series, recommended for those who enjoy logical reasoning and exploration fantasy.
Last week of the first month of new $dayjob after more than half year burnout recovery. I am still way more tired than I’d like, but people are super nice so far and that sure helps.
For $fun, I am trying to subsample periodic signal in a pseudo differential fashion with interleaving that induces a phase offset. I really need to simulate the approach with scipy before trying things out in C.
Hacking on my side project and developing a small open source HTML generation library alongside it. Using it in my backend. No weird templating languages needed!
Trying to make a first release for my pet project Spatial Shell. The code is ready, the man pages are ready, I think i just need to come up with a nice blogpost giving a tour of the project in a visual way to call it a day. What’s nice is that it is a project I have been using daily for almost six months now, which is always super neat (having a project driven by a very concrete need, rather than a fun ideas as some of my previous hobbies were).