It is starting to feel like I am going to have to learn all the theory behind Hit-and-run samplers, and other MCMC / probabilistic programming stuff.

Had a PR merged without further changes, after 5 years github.com/queryverse/Q...

was thinking back about my first vaguely significant open source PR today. Almost a decade ago. Feels much longer. github.com/JuliaLang/ju...

My employer did a spotlight on me for their social media. 😊😊😊 www.linkedin.com/posts/meet-o...

Something I have been thinking about on and off is about the annotation of function primatives with properties, like nilpotent, associative, commutative, etc, then using (type) inference to infer this about functions built in them. Then using that for automatic parallelism and other optimisations.

I accidentally turned on AI completion in VS-Code yesterday. And I though "_fine_, I will finally try this thing" and I am really not liking having my tab completion lie to me? I just have to check it when it doesn't work, then look up the right way. When I could have just immediately looked it up

the amount of hatred i have for good tests that were marked "skip" because they are too slow that have bit rotted, is only exceeded by my hatred for good tests that were marked "skip" that are still good and would have caught bugs had they not been set to skip

I've just created a #JuliaLang package BlueskyClient.jl github.com/AtelierArith... . You can post text, images with ALT text, GIFs, and MP4 videos from the package. Give it a try!

this is a really cool page on all the tricks you can use to express different ideas in terms of (Mixed Integer) linear programming. Cos the whole thing is very tricky. Like the trick of splitting variables so you can have L1 regularisation which is nonlinear in naive implementation

I have spent far too much time dealing with flakey tests and nondeterministic failures today

the amount of hatred i have for good tests that were marked "skip" because they are too slow that have bit rotted, is only exceeded by my hatred for good tests that were marked "skip" that are still good and would have caught bugs had they not been set to skip

yes, I find i often in python I have modules that are really just one function (at least in their public API). Because it makes sense for all that to be in one file. Combined with explict imports and it makes quiet a complexity of import headers.

does python 3 have an abstract base class for strings? I feel like it should. because for data-structures like a rope you want the string API, but you definately do not want to store it like a string, so inheriting from str is no good. en.wikipedia.org/wiki/Rope_(d...

I wanted to see if I was part of the Anthropic settlement class: ✅ Book on LibGen before 2021 ✅ Book has ISBN ❎ were registered with the United States Copyright Office Alas, I am not. While my rights have obviously been violated, it is not in a way that is covered by this court case.

So we had a model incorporating thermodynamic constrains called a ThermoModel. And another model with enzyme constraints called a GeckoModel. (Gecko = Genome-scale model Enzyme Constraints, using Kinetics and Omics.) And we were making a model to combine them. Some kind of ThermoGecko.

I need to set julia to always run with bounds checks enabled. I have done this before. The big virtue is no more redoing precompilation between running normally and running tests. Since julia makes seperate compliation caches for the 3 different settings for bounds checks. +tests always have them on

I made an OpenSCAD library for engraving emoji into 3D prints. github.com/oxinabox/ope...

Pretty happy with this for something I designed and printed all in one day, while doing other things. It snugly fits the brace on the back of mirrors. Which has a funny bumpy shape. No more bracelets scattered over my dressing table.

🚨 Breaking news 🚨 Two epic tutorials announced for #JuliaCon Local Paris 2025: one about Pluto notebooks and one about using JuMP for optimization. Be sure to get your tickets this week, otherwise the early bird pricing will be gone 🦜 @juliacon.bsky.social @julialang.org @pydataparis.bsky.social

I ordered a 3D printer. I haven't had access to one since I left uni in 2018. And that one was ancient and temperamental.

I should implement the same design o want to 3D print in OpenSCAD, Descartes.jl, ConstructiveGeometry.jl, ImplicitCAD and a few other programatic CADs. Then compare and contrast

One julia thing I am coming around to after using python for a bit is explicit imports. I used to be moderately hard against it. Working out where something is from is a tooling problem that is kinda solved. But when jumping into a codebase where you are unfamiliar with the whole ecosystem...

The 1 module per file pattern is not idiomatic Julia. I am dubious of its success in python where it is built in and required. You end up with a hell of avoiding circular dependencies. And trying to match API organisation (external) to development organisation (internal) is just messy. Long files.

Sometimes it's fun to pick up random open source repos and contribute to them. Yesterday I picked up ConstructiveGeometry.jl Which is a Julia based OpenSCAD alternative. It's a 1 person effort not touched in 4 years, and I got it working with latest releases of its dependencies etc. and made a PR

DataStructures.jl v0.19 has been released.. Our first breaking release in 5 years. It shouldn't be too breaking if you kept up with removing deprecated calls. (For reference 0.18 has 22 patch releases which was an annoying amount of backporting to do. Might be a record for backporting) #julialang

Biology be like: So 1 gene encodes for 1 protein right? It's 1-1. Well no, some genes are no coding. Ok so it's 1-0/1 Well no, each AA can come from multiple different codons so two different genes can technically make the same protein. Ok so many-0/1. (It gets worse though)

I actually quite like SQL. Nice language to express things in. Though how different versions of the same query will perform is mysterious

Biology be like: So 1 gene encodes for 1 protein right? It's 1-1. Well no, some genes are no coding. Ok so it's 1-0/1 Well no, each AA can come from multiple different codons so two different genes can technically make the same protein. Ok so many-0/1. (It gets worse though)

Something i just said seriously at work: "How different is a human from yeast anyway? Probably not that different."

What database normalisation rule is broken by having redundant information For example have a column for middle name and another column for middle initial. It clearly breaks some rule as they could be updated separately and end up inconsistent.

I was away for one week and someone added a `utils` file to the code base. `utils` files are an antipattern because it's a meaningless name, an unclear grouping. (A utils folder is ok if the files have good names). And it tends to be where evil code lives. often without tests. Cos "its just utils".

I wish github had a thing that would let me right click an item in a check-list and open a subissue with it's text as the title. And edit the original post's checklist to be a link to that subissue. That would make breaking stuff down for project management purposes much better.

I thought Dask was supposed to be this really reliable thing, especially when using it with Coiled so I can't screw up my infrastructure. But I have been fighting it all day, of it seemingly getting to a point where it stops talking back to client...

NetworkX allowing anything to be the nodes in graphs is solidly nicer than Graphs.jl Look how nice it is to go from a CobraPy model (a reaction network) to a (bipartite) Graph:

Python is a 1 based language. It's just that indexing is written as `x[-i]`. The hyphen is important! Also lists are printed backwards.

Classic bio-data hell 🥰🥰🥰 I realised our tests depend on 5 subtly different GEMs for the ecoli core metabolism. And they can't be substituted for each other without some rework. Plus 2 for yeast, and 1 for salmonella.

for any axis L of a convex polyhedron H a function taking a point in P and returning the cross sectional area of a slice of H is convex. similar for any direction D perpendicular to L, a function f_D(p in L) that returns the width of H in that direction D at point p, is convex. So a question:

Inheritance was a mistake. Multiple inheritance, multiply so.

I reviewed #3000

JOSS has now published more than 3,000 articles, and I've personally edited 60 of them (still editing +5). Being a part of JOSS has been one of the most fulfilling experiences of my life. @joss-openjournals.bsky.social #JOSS

Anyone else with Github Teams suddenly see all Github Actions failing because if billing reasons? Today we started see that all of a sudden and i was confused cos out billing looks good. Then i noticed we had a new(?) budget tab. Which said 100% used ($0/$0). Changing that got things working again

The coworking space definitely made the phone booths uncomfortable on purpose to stop people staying in them too long. Stool rather than a chair. No power outlet. Tiny shelf of a bench rather than a desk. Very small with bad airflow. Similarly the kitchen area has stools.

A thing about being at a very small startup, a good nerdsnipe can absolutely sink total productivity for the day. I took out both founders today with a weird Uniprot entry.

I am now working from a coworking space. I spent a bunch of time yesterday/today finding a good one, and I picked the best. Ergonomic chairs and monitors provided. But one thing I do not like is that they have managed to get the vibe of a basement even though it's 1-2 stories above the ground.

@metanetx.bsky.social what is the ETA on v4.5?

A new metanetx release is overdue. I am looking forward to it because it will fix some bugs I found. But also I fear it because it will create new ones and void some of my local fixes

I have been in Melbourne this week on boarding three new staff. Due to coincidences, for their first 3 days the CEO was at a conference, and the head of Eng is in UK for a few months completing a degree. So it would have just been our admin person. So I flew over from Perth. I ❤️ soft-skills work.

Should I switch to always `git rebase --update-refs`? (using the global config) dev.to/onepoint/git...

Em-Dash is capital underscore. Underscore is lower-case em-dash.

I suspect the more standard thing to do would be to build the docs, push them to S3 with github.com/marketplace/... and use AWS cloudfront to setup some basic auth (like even just a shared password for whole company would be fine. Its only docs)