Hacker News Reader

Comment by ShabbosGoy | Hacker News Reader

> Not only spot, but also invests the energy in calling it out and explaining it.

Which is why FOSS is so important. Can you imagine a world where Linux was closed source? You’d never have any insight into security patches, or these discussions.

View on HN

> FOSS is so important

FOSS is nice, but it's not a silver bullet. In this case I think Linus is largely to thank. Things will be a lot harder if he decides to retire.

A silver bullet is having one smart guy at the top who doesn't care how much pressure you exert on getting something into the codebase, or how much man-hours were spent (sunk cost fallacy). I think in a lot of "corporate" projects (not-open source) there's much less of that and the software is designed by commission, with the lead developer(s) quickly no longer having much of an influence on the code itself, being drawn away into meetings and such.

Maybe I'm projecting.

It’s an ideological situation to say the least... but it’s also what we have right know and god damn I’m thankful for it’s beautiful rarity.

> A silver bullet is having one smart guy at the top who doesn't care how much pressure you exert on getting something into the codebase,

Well if Linus goes too far, Google and a couple other big corporate contributors could team up and fork Linux, and it would be very hard for the open project to compete. I think the fact that Linus is so good is a big part of what makes the whole thing hold together.

They could, and they often do (Android and ChromeOS effectively run forked kernels). At the end of the day though, they always wanna rebase against that sweet sweet mainline quality.

It's simple, the stack of hacks they produced in the meantime at some point becomes a tall tower prone to fall over.

I think this is dead on. I think at Microsoft that guy was Dave Cutler. Most of the really well designed stuff seems to have him at the helm - NT4.0, bits of XP, Win2k, Windows Server 2003 and then Azure.

I have a suspicion that Satya's rise to CEOdom via Azure's success could be attributed to him too.

If you're a Cutler fan you may enjoy Showstopper[1], an account of the creation of NT.

1: https://www.amazon.com/Show-Stopper-Breakneck-Generation-Mic...

Thanks for the rec, looks interesting.

It's a great book. Motivated me to switch industries and relocate to Seattle as an NT kernel programmer.

This may be true, but FOSS does not magically make it happen. Many FOSS projects are led or dominated by large companies today; the development may happen in the open, but it is as "corporate" as it gets, and no bullshit calling occurs (or if it does, it is quickly silenced with EWONTFIX & co.)

That's only partially true. When you only have one corporations interests, things get shady very quickly. When you have multiple, competing corporations, they have a vested interest in keeping their competitors honest. It would be very rare indeed to find something that every company wanted to downplay at the same time.

Spectre and meltdown are a great example of this. Sure, ARM, Intel, and (to a lesser extent) AMD may want to downplay these things, but Google did not, so it became a big deal anyway.

> This may be true, but FOSS does not magically make it happen

So what? Nobody said that. This is a strawman.

Of course FOSS is not sufficient. Look at any homework-style GitHub repo. Nobody disputes that.

All that is said is that FOSS is necessary to achieve this level of quality.

We may be we can train an MBA to do Linus's job? ;)

So you demote FOSS from "important" to "nice" just because it isn't a silver bullet? (which does not exist, anyway)

That doesn't make any sense. This is a strawman argument.

Nobody ever said that FOSS is a silver bullet, not even FSFE, FSF, or Stallman himself.

This is as good a place to ask the question nobody likes to discuss, but is important to never forget - what do we do when the bus hits Linus?

I believe Greg Kroah-Hartman is the designated successor.

In most other FOSS projects people would be afraid to be that frank when referring to Intel's behavior though.

Exactly, I’d love to see the state and flux of the Meltdown and Spectre patches for the various Windows kernels.

Worse still, imagine a world where there was no Linux, and Oracle/Sun happened?

Ironically, Sun might have not had to sell themselves to Oracle, and the world's server's might be running on SPARC chips that are immune to these issues.

Linux is what allowed Intel to get its biggest foothold into the UNIX server market.

This is true, but if there hadn't been Linux there would have been something else to fill this niche (runs on commodity PC hardware, zero license cost). Sparc/Solaris wouldn't have continued to rule. Intel would have found some other workable solution.

Indeed. For one there was the 386BSD lawsuit, for another there was HURD. Torvalds himself has stated that if HURD had been usable at the time, he would not have bothered working on Linux.

Honestly, back when Linux started getting momentum, RMS' GNU/Linux argument had merit. Most of the stuff people where running on top of Linux, never mind using to develop Linux, came from the GNU project.

True. And we'd have the same issues with this processor family now...

Don't get me wrong, I love Linux and loved FreeBSD back when it was more easy to choose... But I do think the 'free UNIX' movement has had unintended consequences like the demise of Sun, and perhaps now also this exposing of a failure in what was a desktop class chip architecture that was elevated to server usage.

Everything has a cost is all I'm saying.

>Ironically, Sun might have not had to sell themselves to Oracle, and the world's server's might be running on SPARC chips that are immune to these issues.

Solaris, with the exceptions of a few bright stars like DTrace and ZFS, always seemed like a POS to me, compared to linux. Especially as a work-station. Sun went full retard with Java everywhere right before they went under and I have a feeling the Solaris Desktop experience, and by extension the UNIX desktop experience, would be something from my worst nightmares, something to make wish for Windows Vista.

> and the world's server's might be running on SPARC chips that are immune to these issues.

Given enough time, SPARC chips would also become vulnerable to these issues. See what happened to ARM: their later (higher performance) designs are more vulnerable than their older designs.

That world was the late 90's / early 2000's. Yes, Linux was around, but every startup was buying Sun hardware since it wasn't quite mature yet. Sun was the dot in .com after all.

I actually worked at a startup that threw out a working Linux / MySQL solution in favor of a Sun / Oracle one. Because, you know, VC...

Linux is a kernel. An x86 Kernel running Gnu would have happened, and instead of Linux we would just be calling it something else.

Reply on news.ycombinator.com