Hacker News new | ask | show | jobs
by seisvelas 2106 days ago
I apologize in advance for such a brazenly ignorant question: But why is it that this small team is able to produce a working microkernel, yet GNU Herd and (more recently) Google's Fuchsia have languished eternally as vaporware?
7 comments

The manager asks the developer "how long is this project going to take you?" and the developer replies "2 weeks." The manager yells "That's unacceptable! How long will it take you if I give you another engineer?" to which the developer says "4 weeks." The manager asks again "Fine, take as many resources as you need. How long if I give you 100 engineers." to which the developer replies "it'll only take us 30 years to finish the design and after that it'll pretty much be smooth sailing!"

jackpot51 describes himself in his bio as a "BDFL of Redox OS" which means he's likely personally passionate and just does what he thinks is best. This will likely not lead to a better outcome than Herd and Fuchsia would lead to but it will lead to an outcome within our lifetimes.

In many cases, however, "working" is better than "planned"

You can use GNU HURD. I did so something like 20 years ago for kicks, then went back to Linux.

The reason stuff doesn't take off is that it never takes off. Implementing an OS is actually not hard. You can get pretty far as a 1-person project provided you have the time. Real world use and other people adopting your stuff is harder.

You can even use gnu/hurd, the way gnu/Linux naming was trying to be specific about. https://www.debian.org/ports/hurd/
I recently found the archives for oskit and gnumach on oskit, but imagine my surprise that it won’t bootstrap lites...

I can’t imagine how difficult it would have been to get support in 94, but 2020?!

By "a working microkernel" you probably mean a working microkernel-based operating system.

If so, besides the hurd (and all the debian packages that build and run for it), there's Minix 3, HelenOS and Genode (specifically Sculpt) to look into.

I don't think fuschia is vaporware. But I think the reason this seems more advanced than Hurd even being a lot younger is (also) because the concepts regarding microkernels have been expanded and solidified a lot since it started.
What makes you think Fuchsia is vaporware? It's not general purpose but google is definitely shipping/going to be shipping devices with it.
There were used to be GNU/Hurd distros and Fuchsia (not a microkernel according to their FAQ; the reasoning being the many syscalls) was bootable with user interface within two years. Hurd development was/is slow because of architectural decisions they took, then Linux took off and GNU developers concentrated on providing good tools for it.
In fairness the project has been going since 2015. Don’t know about Herd but Fuchsia is a lot more than a microkernel, right?
Fair points! But,

>the project has been going since 2015

True, but Fuchsia started only a year later

>Fuchsia is a lot more than a microkernel, right?

Indeed, but that is also true of Redox!