Vinix – An effort to write a modern, fast, and interesting operating system in V

I'm always interested to hear about people developing OS's. Hope they implement some stuff other ones don't have.

I never heard of V so I looked it up:

https://vlang.io/

An implementation of 2048:

https://github.com/vlang/v/blob/master/examples/2048/2048.v in V

I remember shenanigans when V was first announced (release date for public preview, but for a little money you get early access), people being upset that it was half-baked, etc. and wrote it off, but I've heard that it has since changed a ton and is nothing like the early days.

Can someone familiar with recent V developments comment on this? It makes bold promises that would be awesome to see realized, but I got the feeling there was just a lot of smoke and mirrors going on in the early days.

This reads as a bit more ambitious than the usual HelloWorld or http framework example project.

Oh also, mintsuki's (limine, lyre, echfs, qword, some things with managarm, etc) comments about V vs C/Rust/Odin/Zig are really interesting in #vinix-os: https://discordapp.com/channels/592103645835821068/645024404...

Things have improved but it's not clear the author realizes the magnitude of what he's trying to do.

The big ticket item is the autofree memory management system which promises near C/Rust performance and minimal overhead without the complexity of borrowing and ownership. To date, this system does not work and is "wip" as the vlang community likes to say. There are also no clear explanations of how it works beyond just "it inserts calls to free() when you are no longer using a variable" which is ... extremely obvious. The more interesting question of how the compiler determines something is no longer being used in the face of aliasing and no borrowing system is completely unanswered. No one working on the project understands how complex that problem is. Initially the goal was that this system would manage all memory for the programmer automatically, this goal has since moved to "manage 90% of objects in memory and use GC for the rest". I suspect we will see that number continue to drop as the author realizes this is intractable without whole program analysis (which they can't do and still hit their compilation time targets).

Other items are very much in an alpha state if they work at all. Only in the last month have you been able to create a generic function with more than one type parameter, although V has been claiming support for generics since at least last year. There's still no ability to constrain the type parameter to something that implements an interface/trait/protocol.

C2V, another wildly ambitious project that could literally be a PhD project in automatic language translation, was supposed to be released "this week" 5 weeks ago. It's currently in a very private alpha and no source has been released.

Which is basically the state of most of the author's projects: huge promises in terms of features and performance with most of them little more than demos. I've come to believe this is an intentional strategy by the author to maximize his Patreon revenue. He makes big promises on social media and then delivers the easy 20% of the project to his followers before switching to another project. It would make far more sense to release a finished project before starting another massive one and yet he can't sit down and finish a project because he knows he can't actually deliver what's been promised.