qrobit

Arguably PCIe lanes the thunderbolt exposes are a differentiating factor, not the maxed-out specs that "plain" usb-c cables can have. Although to be fair, I've never used the PCIe lanes on thunderbolt ports.

And it comes with tidyverse-like cheatsheet[1] that I confused with ggplot2 when first discovered plotnine

[1] https://github.com/rstudio/cheatsheets/blob/main/plotnine.pd...

Sort of, there're more posts about graphene in the year 2026 & they get much more attention. Aggregated some data and plotted it with my agent: https://boop.icu/Pr.png

That's what I wondered as well, whether it is a new implementation, or perhaps they still use their version of cups? Or maybe they use the forked openprinting cups? Would be nice if somebody knowledgeable with a mac device told whether the cups is still used

The repo got archived in march, I wonder whether it means that Apple uses something other than CUPS on their devices? Quick search suggests they still use CUPS, but why would they archive the project then?

previously https://news.ycombinator.com/item?id=46906947

On nixos they're not, there's a single executable in /bin -- /bin/sh, and a single executable in /usr/bin -- /usr/bin/env

Both are non-deterministic, both have some metric to optimise, one is specific and efficient, the other is too broad and very expensive

Much older? Wikipedia says[^1][^2] java appeared in 1995 (started in 1991), while python appeared in 1991 (started in late 1980s). 4 years doesn't seem too far apart, considering both language are >30 years old now.

[^1]: https://en.wikipedia.org/wiki/Python_(programming_language)

[^2]: https://en.wikipedia.org/wiki/Java_(programming_language)

He doesn't have a role at Facebook anymore AFAIK, although I can't find the source now. I remember hearing it was mostly due to filtering engine at Facebook being rewritten in Hack, so they don't have the need for Haskell people anymore.

Secure boot ensures the image you boot was not tampered with. You can't install keylogger without tampering with the image. If you wanted to install physical keylogger, you would need to open the device up, and at least my laptop provides detection of bottom cover removal, meaning the system will ask you for a bios password if the laptop was opened up.

you don't have a colourblind person's 'profile'

Why not? As far as my understanding of color blindness goes, you just need to find a precise transformation matrix and offsets to be able to correct any type of deficiency (except for achromatopsia, I guess).

When I use them I find the colour profile I am used to in the wider world flipped, and the semantic meanings given to colours, or their hierarchies, completely changed.

I think the correction applied to digital content is a positive thing. At once you can perceive color the way it was intended to be perceived. May be wrong here because I don't have daltonism.

You can try to make a perfect system for every variation but the end user won't see it as precisely as you intend.

My goal is not cover every case, but to create exactly one profile (and perhaps create a usable correction workflow for someone else)

Thanks for an elaborate response :)

There should still be privacy concerns, especially with their demo which sends a POST on "Generate". The author suggests the model is 85kB of weights, which could run perfectly well in browser.

But with this software - the tolerances are looser, so the clothing becomes more manufacturable.

Does it? How do looser measurements help? I assume manufacturer would always take the upper bound of dimensions. Suppose model also predicted your dimensions are higher then they really are, so these two in combination give you an oversized piece of clothing.

I personally like marimo[^1], the developers rely heavily on AI tools afaict

[^1]: https://github.com/marimo-team/marimo

Just a reminder that even if you managed to amend those commits and force-push, the commits would still exist and will be addressable given the hash is known.

Hackage recommends using Haskell's PVP[^1], but does not enforce it. That's why many haskell packages are a four-places versions: 3 required and fourth optional (but popular) that represents "other" changes, like documentation.

[^1]: https://pvp.haskell.org/

Mostly true, but I personally have it turned of for duckduckgo and it shows me some ads with [ad] label. Actually if you wanted to disable ads there, you wouldn't even need an ad blocker, there's toggle in the settings

It will be easy to check with devtools when the update is released

I was surprised not to find cantata[1], another MPD graphical client, on the list. Used it for the past three years, despite it being unmaintained for quite some time now. The client is very featureful, allows downloading lyrics and covers automatically (TBF had many mismatches, like downloading some Gillette ad as an Eminem's album cover). Most important to me is the ability to listen do directories and not artists/albums, which cantata does perfectly. Recently nixpkgs replaced cantata with a fork[2], so cantata is kind of online again.

[1] https://github.com/CDrummond/cantata

[2] https://github.com/nullobsi/cantata

Bubblewrap seems to be much more popular[^1], personally this is the first time I heard about landrun

[1] https://repology.org/project/bubblewrap/information https://repology.org/project/landrun/information

I wanted to say Haskell with shh[^1] and stack's or nix's shebangs[^2][^3], but interpreted haskell is not particularly fast.

Also I think a Python script is reasonable if you use a type-checker with full type annotations, although they are not a silver bullet. For most scripts I use fish, which is my preferred interactive shell too.

[1] https://hackage.haskell.org/package/shh

[2] https://docs.haskellstack.org/en/v3.9.1/topics/scripts/

[3] https://wiki.nixos.org/wiki/Nix-shell_shebang. On a side note, if you were to use nix's shebang for haskell scripts with dependencies, you should be using https://github.com/tomberek/- instead of impure inputs, because it allows for cached evaluation. I personally cloned the repo to my personal gitlab account, since it's small and should never change

I would argue that the barrier to entry is on par with python for a person with no experience, but you need much more time with Haskell to become proficient in it. In python, on the other hand, you can learn the basics and these will get you pretty far

I think smalltalks just didn't bring you to the right topic, if you were to reach topic you both fancy, you would definitely remember each other. If anything, having a smalltalk is nicer than staying in awkward silence.

Unless there is some vulnerability in the current version that you want to take advantage of. See e.g. mediatek exploit to unlock bootloaders without authorization by OEM or hacking PS4.

Great post! Reminds me of (UK) Passport Application game: https://jameshaydon.github.io/passport/

While I don't use GitHub for personal projects, I don't see how moving off GitHub solves anything. If you are open source contributor, then repositories you interact with are public, and there is nothing preventing anybody (including GitHub) from using your code for any purpose. If you contribute to a public repository on any public platform, be it Codeberg, sourcehut or GitLab, your activity is public, issue you create are public[1], everybody knows who changed what in the code.

The social media effects are twofold, on one hand I think stars and contributors count, contributor profiles are great to see what is popular, but it went a little too far when they added scrollable home page. Virtually everyone has an account on GitHub, the best way to make your project visible and ease the contribution threshold is to put your project on GitHub.

What I would like to see is federated git, so that some protocol allowed different git servers could communicate with each other, which will make moving off GitHub much easier.

[1]: except for sourcehut I guess, which does not have issues or pull requests?

As a younger person myself, this talk is incredibly entertaining, kind of eye-opening too; the passion for engineering that speaker radiates is immeasurable.

I suppose because modern websites are actually apps, in an app you click on stuff to see more content

If web page does not look like a blog or a newsletter, it is not wrong to assume the app format

For me lenses are like water. I start each haskell project with `optics` dependency, my favorite trick is deriving Generic and using labels to access the fields[1]. Field accessors never get created in the first place, so you have no namespace pollution and all your data definitions can have fields called `id`, `date`, whatever and never collide with each other or imports from other modules. The only problem is unwieldy error messages in case something goes wrong. Another tool is Magnify and Zoom[2] classes, which are useful for limiting context available to child actions in a monad stack. You can have one "global" state, but give only a slice of that global state to the child action with `magnify #subState1 (myAction1 arg1 arg2)` (or zoom respectively for MonadState actions). This approach is extensively used within Brick[3] TUI library, where you have big "global" app state, and a collection of widgets that have their separate states which you slice from the big state.

[1]: <https://www.youtube.com/live/9hrDm7xDpig?t=2200>

[2]: <https://hackage.haskell.org/package/optics-extra-0.4.2.1/doc...>

[3]: <https://hackage.haskell.org/package/brick>

FWIW, these uBlock Origin rules solve the ploblem:

    ||accounts.google.com/gsi/iframe
    ##iframe[src^="http://accounts.google.com/gsi/iframe"]
    ##iframe[src^="https://accounts.google.com/gsi/iframe"]
    ##iframe[src^="//accounts.google.com/gsi/iframe"]
    ###credential_picker_container