porgamrer

You have imposter syndrome so bad you don't even think you're a real imposter!

That sounds plausible. I'm sure having the lowest possible latency was their goal. There are multiple popular rust libraries aiming to provide zero cost abstractions over a common subset of the metal, vulcan and dx12 APIs, but I've never actually used one.

But why? I saw their section about rasterising everything on the GPU, and again I find it hard to understand why they need anything more than OpenGL 3.

Their UI looks like it would be an LRU glyph cache and a sprite batcher, and then you'd have 1000fps.

Thanks for clarifying. Do you know what the bug is? I gathering it's something to do with enums and boxing to enable a "safe" transmute, but couldn't make sense of the code.

Scary compiler edge cases with value-type enums aren't surprising though. Trying to implement that feature with memory safety and high performance sounds like a nightmare.

I'm still confused by this not being cross-platform. It's made in Rust; basically every graphics library is cross platform out of the box, and so is all the file IO stuff. There will be some specialist OS api stuff in places but surely it can't be much.

For once this comment isn't even snark. I acknowledge my ignorance and wonder if someone could explain why the cost is bigger than I think?

Perhaps it's setting up CI and packaging for other platforms? Maybe they want human QA on every release? Maybe the APIs for slick OS integration are more complicated than I realise? (e.g. putting UI in the taskbar)

Counter-point: Atom is terrible. Its electron competitors are terrible. Big IDEs are terrible. Simple text editors are terrible.

If you are under 50 and chose to learn vim or emacs, there is a 100% chance that you were also forced to learn latin at school and honestly it's not your fault that you turned out this way.

These are all the options. Sometimes all the options are terrible.

Here's one I learned from a past manager:

Stress that everyone needs to pitch in and make themselves useful at all times, but do not share any information at all

Make sure the work is not broken down into clear tasks. Make sure nobody else has access to the stakeholders. Make people ask separately for every single account or access credential they need, and respond with incredulity that they don't already have it.

Give the impression that there are no processes. When someone submits work, criticise them for not following the process.

Each day, schedule meetings so you are impossible to contact until the early afternoon. That way you can interrupt any request for information by asking the person what work they did in the morning. The goal is to close the loop by making people scared to talk to you, so they blame themselves for not knowing anything.

I don't need this level of introspection on a tuesday morning

Rewriting bits of the kernel makes sense. I can't imagine them porting much C# to Rust though, beyond very small, self-contained services.

Everyone likes a dramatic headline, but in my estimation there is 0% chance of Microsoft pushing widespread Rust adoption over C#.

In the long-term I'd guess they are more likely to continue extending C# with features that make it possible to optimise hot loops. They already added NativeAoT and ref structs, and they have done a lot of research into memory regions and capabilities (an alternative to Rust's affine types).

Eventually it may be possible to opt into a clunky language subset that gives Rust-like performance without giving up memory safety.

They are also quite likely to use OS-level intervention to safely sandbox C++ code inside a .NET process without giving up performance. They've done a lot of research on this, and now they can steal notes from webassembly too.

Haskell has very famously not solved this problem. As in, pretty much every academic paper published on the subject in the past 15 years includes a brief paragraph explaining why Haskell's effect monads are terrible.

Also, it would be surprising for Rust's developers to be scared of monads when Rust already has monads as a core language feature, with special syntax support and everything.

Or adding a single non-corkscrew pipe out of principle, which all the other corkscrew plumbers now have to maintain for 20 years

I wonder what the planned scope is. C++ is such a fundamentally uncooperative language.