MirageOS Pioneer Projects | Hacker News Reader

MirageOS Pioneer Projects

github.com

55 pointszem14 comments

If any of these prompt you to think of new potential projects that's cool too. Just email the list to discuss it.

Also, a few folks raised a question about what the difficulty levels actually mean [1]. If you're new to OCaml then the suggestion is that you pick an area you're already familiar/comfortable with (eg networking, Xen, etc) so that you have some grounding. Otherwise, you're having to learn two things at once.

[1] http://openmirage.org/wiki/weekly-2015-02-25#PioneerProjects

MirageOS is already a thing, I wish they'd pick a new name.

http://www.ticalc.org/archives/files/fileinfo/139/13949.html

Reading the title, I thought it was talking about the calculator shell too!

Their website is still up: http://www.detachedsolutions.com/mirageos/

From the ticalc.org page, "Ranked as 2 on our all-time top downloads list with 1003409 downloads."

Over a million downloads over its lifetime, that's incredible. I never realized just how large the calculator homebrew scene was in its heyday.

It's a pity all this effort is going into Ocaml which seems something of a dead end.

If the core language was Rust or Go then momentum would really pick up.

OCaml's been around for 30 years, and will be around for many decades more. It's not this year's fad.

It's not about fad. It's about providing unikernel technology to the mainstream development world where it's actually needed. Mirageos is not going to drive uptake of ocaml in any big way. It makes more sense to give the people the tools they need to get the job done, like some less arcane language than ocaml. As you point out, it's had 30 years to catch on but hasn't.

> "Mirageos is not going to drive uptake of ocaml in any big way"

Right, because Unix had no effect on the number of people working in C.

> "It makes more sense to give the people the tools they need to get the job done"

MirageOS takes a clean-slate approach and it's not exactly a secret. There are evidently significant benefits to this, as well as trade-offs. There are other unikernel implementations out there that have prioritised legacy over safety and composability.

You mention Go as an alternative. I'm curious how Go is any less arcane.

http://www.quora.com/Do-you-feel-that-golang-is-ugly/answer/...

> less arcane language than ocaml

There is nothing particularly arcane about OCaml. The fact that it's not a mainstream language is not a good indicator of its complexity.

> "... which seems something of a dead end."

Based on what, exactly? If anything 'all this effort' should indicate the complete opposite of what you state.

Unikernels won't take off till they can be built with languages that aren't ocaml. Sorry if that's not a message you're comfortable with but its there.

It seems to me that you haven't looked into this space at all. See Table 2 in [1] for a list of other implementations.

I'm perfectly happy about how MirageOS is doing and I'm glad there are options that allow people to choose where they make their trade-offs.

[1] http://queue.acm.org/detail.cfm?id=2566628

I guess all that's left to do is for you to start your own Unikernel project using a technology you feel is more appropriate.

Reply on news.ycombinator.com