hinnisdael

Agree about Voxtral being the only model — local or cloud — that handles multilingual conversation really well. I‘m not sure what they do differently, but mixed-language sentences and industry terms don‘t seem to faze it where other model begin to struggle.

Sounds like you might be better off freelancing. With a bit of networking and/or marketing you‘ll find individual clients willing to pay for your perspective on their projects. It‘s a different thing with outright hiring — as a manager I‘d find it hard to justify picking you over someone more classicly „predictable“. Most companies do care more about predictability then about spriritual compatibility in the end.

The project presents the sounds of 270 UNESCO World Heritage sites and items of intangible heritage – to create a brand new way of experiencing these spaces.

> You won't find any […] active safety systems […], nor ABS, nor airbags.

I‘m surprised this is legal to drive anywhere without airbags, whether they‘re targeting Southeast Asia or not.

Agreed, it‘s great microsite to showcase the font. Another one that I stumbled upon lately is MD IO - a great font and a great minisite as well: https://io.mass-driver.com

What a great font! Very matter-of-fact, but with some stylistic choices that give it just enough character to be interesting. The x-height in the original version is too large for setting body copy, but the variable font comes with an axis for the optical size, which allows tweaking the x-height individually.

Yeah, the audiobook version really is quite special. It’s read by Peter Vilnai, if I remember correctly. One of the few audiobooks I was able to completely be absorbed by while listening.

The water is fine, but the sink usually isn‘t. The sinks are just as contaminated with bacteria as the toilets themselves —- sinks in really busy restrooms are constantly damp and allow bacteria to grow more than one would expect.

Great advice: describing things in order of importance.

Most people intuitively describe images from foreground to background or left to right, a bit like they are mentally completing a checklist of all the things to describe. As correctly noted by the author, describing by importance first has the added benefit of allowing screen reader users to skip irrelevant/uninteresting images early.

Compare:

Torn-up painting in a gallery, observers standing in front of the work.

vs.

Gallery interior, people standing in front of a painting with visible damage.

I would argue you no longer need a framework when using CSS grid. Creating template areas once per breakpoint and assigning them once per container really is all there‘s to it. Sure, you can build another abstraction on top of that, but I doubt it‘ll get any simpler than the underlying handful of grid declarations.

Aren‘t you losing information in the parts that aren‘t perfectly straight? Yes, you can stretch those to recreate the original layout, but that would come at the cost of resolution in the interpolated sections of the page. Granted, not a problem for most books, but probably a reason prople are still looking for mechanical solutions to the problem.

Right, no public pricing usually means rates far above what personal or low-budget projects can afford.

Not sure how much of the design is protected and how much inspiration one can take for a non-commercial DIY project like the one presented by OP.

There‘s a somewhat similar commercial scanner [1] [2], with a V-design as well but inverted to scan from the top. Much gentler on the books as it‘s the scanner that moves, not the books themselves. Super happy to see someone develop an open-source alternative!

[1] https://www.treventus.com [2] https://youtu.be/SdipuAuWsEs?si=dFWRtva5gO2oM91o

Swup maintainer here. There is official support for parallel animations in swup v4: https://github.com/swup/parallel-plugin — we've been holding off on this for a while because of accessibility issues and some technical complexity around animating those. But we're very happy to have a first-party solution for this now: some UI interactions just make much more sense when still having the previous content around.

vidstack is very light on technical details but starts with a concise intro and a screenshot, as well as relevant links: https://github.com/vidstack/player

payload is well-structured in general: https://github.com/payloadcms/payload

nanostores starts out with an intro and telling code examples, followed by lots of technical details: https://github.com/nanostores/nanostores

I'm currently converting a suite of E2E tests from Cypress to Playwright. In hindsight, I'm surprised by how much more idiomatic the Cypress code was in comparison. The concepts just make sense, whereas Playwright is very unintuitive in places. Cypress seems a bit like the jQuery of browser testing. It just works without too many surprises.

What's really fantastic about Playwright is the automated HTML reports it spits out.

I like what this is trying to accomplish, however I'm somewhat confused about the immediate goal here. Is this for learning about services? Is this about assembling a tech stack? And how does the example site fit in that's mentioned multiple types, Prompts with Friends? This probably needs slightly better wording in the first one or two paragraphs.