Redox OS Making Progress On RISC-V Support, Eyes QEMU & Neovim Ports

[HEX0]

no thanks

until i hold a device, it remains theoretical. i remember spending many hours trying to find an L4 microkernel OS to play with, and never found one.

meanwhile redox and helios just built their own. wonder why...........................................

L3 was was to test if efficient IPC could make micro kernel fast. L4 was then rewritten in assembly and was supposedly 20 times faster than Mach kernel.

Later L4 was rewritten in C++ to see if high level language could still achieve the high performance.

Drew Devault said that Helios exist to prove a point that Hare is suitable to write kernels. Once you create the language, next step is to write a kernel in it to prove said language.

https://archive.fosdem.org/2023/schedule/event/helios/

Similarly Redox OS is to test viability of Rust for writing kernel and the rest of OS.


Now why would either of the projects use L4 written in C++ when the whole point of Redox OS and Helios is to test the limits of Rust and Hare for OS programming.

[mos87]

The more they "eye" the sooner they'll fold. So all good.

[dragon321]

See!
When the eunuchs create their own kernel, it's all good and there is no conflict.

There is no conflict in Linux either, just few people that refuse to admit that the world moved on and modern developers are not going to use low level languages like in the 90's.

[Almindor]

I tried the latest release and found some odd filesystem issues when porting a Rust CLI program. The scary part is that it wasn't consistent. I think Redox needs some more time to mature on the basics.

[Quackdoc]

I tried the latest release and found some odd filesystem issues when porting a Rust CLI program. The scary part is that it wasn't consistent. I think Redox needs some more time to mature on the basics.

ofc it's not consistent, it's not the same os.

[Almindor]

ofc it's not consistent, it's not the same os.

You don't understand. I mean it was inconsistent on the odd behavior. I debugged it down to a call to `mkdir` via the relibc (from a Rust stdlib call). It only failed in specific conditions of the app and FS structure. I asked on the chat and apparently there's some relibc issue.

I'm not trying to be overly critical just saying that it seems even the basics are still very buggy on Redox atm.

[Quackdoc]

You don't understand. I mean it was inconsistent on the odd behavior. I debugged it down to a call to `mkdir` via the relibc (from a Rust stdlib call). It only failed in specific conditions of the app and FS structure. I asked on the chat and apparently there's some relibc issue.

I'm not trying to be overly critical just saying that it seems even the basics are still very buggy on Redox atm.

oh interesting. i've tried to tinker with relibc a bit, hadnt had any issues like this myself, but I tried it for a total of 5mins with a single app xD

[KernelCrasher]

it seems even the basics are still very buggy on Redox atm.

Of course, and it's not something unexpected given that the project did not declare any stable releases.

I'm keeping my fingers crossed for them to keep developing it. It's an interesting experiment. We need people who try new avenues to bring some real progress and not stick to beaten paths far more than we need whiners, naysayers, haters and old farts.