Western Digital Proposes Zonefs File-System For Linux 5.6

[starshipeleven]

You make a big assumption that this code will not interact with anything else in the Linux ecosystem.

No I actually know what I'm talking about, to the contrary of you that start a rant about politics and bs.

Filesystems and other hardware support modules have been merged quicker than other features because it is a completely self-contained module that works only on specific hardware (or on specific filesystem). As long as it does not look like complete garbage (like Realtek code drops for their USB wifi chipsets)

[profoundWHALE]

That would to be fun to read in the LKML. Linus gets egg on his face only to admit that he smacked himself in the face with that egg.

I've read what you've wrote like 5 times and I still don't understand it

[NotMine999]

I've read what you've wrote like 5 times and I still don't understand it

Start with the post from starshipelevenThat post correctly points out that it's a filesystem and you don't have to use it.

I say that Linus is setting a bad example by merging non-critical (you don't have to use it, right?) code that is untested in "staging". There is no imperative to merge into mainline any non-critical code that is untested in "staging".

If zonefs starts blowing up on users, Linus owns the PR problem and nobody else. That's an "unforced error" by Linus which can be avoided by following his own code maturation rules; code should go through "staging" before going into mainline.

But what if the zonefs filesystem proves to be better than most other filesystems and it develops a vocal and loyal following? btrfs is certainly like that.

Then the zonefs filesystem starts to develop odd problems that might be difficult to track down? Some in these forums can share better than I the issues that btrfs has had to deal with. Some of those issues might have gotten it demoted from consideration as the "primary FS of choice" in Fedora.

Even Micro$haft is having issues with Windows Search in Windows 10. A recent mysterious bug has prevented it from SEARCHING. wtf! Ok, it's a feature that you don't have to use, but many Windows users find it useful, and that borkage has the Windows online community up in arms. Even long-time and highly experienced Windows reviewers are calling B$ on Micro$haft's claims as to the cause of the problems. What's more interesting is the mysterious way in which this "borkage" quietly appeared in user's systems; it was not packaged in a Windows Update.

Fact: When you can simply push stuff out to "production" without testing, then crazy borkage can happen. Why take the risk?

That's why I think Linus has set a bad example by merging non-critical code that is untested in "staging".

[computerquip]

You make a big assumption that this code will not interact with anything else in the Linux ecosystem.

Why not be safe than sorry and follow Linus' own established rules for new features; test it in "staging" for at least 1 release? What is the big rush to mainline this code? Nobody will die tomorrow, the Earth will not stop rotating or even orbiting the Sun tomorrow if the Linux kernel does not have this code. Or perhaps you are a "flat Earther"?

Seriously, Linus has railed in the past on others IN PUBLIC for not soaking new code in staging. Remember the KDBUS incident and Linus' railing on GK-H over that? Yet in this case Linus goes about tossing his own rules out the window for himself. Granted, that's an apples & oranges comparison right there, but the point is simple, "The same rules should apply to all involved in the project when it comes to new features. No exceptions."

Or perhaps you would prefer that some (Linus in this case) ought to be more equal that others?

As the top-level code maintainer you can argue that Linus IS more equal than others, but that suggests he could show less regard for the project's "new features" rules whenever he pleases rather than behave in the same way that he demands of others in the project. Said another way, "Do as I say, not as I do." Are you ok with that sort of attitude?

Where is the good leadership concept of "Lead by example" in this case? Good leaders demand of themselves what they demand of others.

So ... if zonefs starts to cause problems for users, then the only person Linus can flip off & rail against regarding those problems would be himself, right?

That would to be fun to read in the LKML. Linus gets egg on his face only to admit that he smacked himself in the face with that egg.

Could you be more dramatic...?

The code is maintained by a separate company who's job it is to QA the code. Because the code is isolated in a module and likely not enabled or built-in by default, there's virtually no risk of this code doing anything to the rest of the kernel unless a user explicitly enables it by compiling their own custom kernel or enabling the module. In addition, the code is about 1600 lines meaning it's small.

[starshipeleven]

Start with the post from starshipelevenThat post correctly points out that it's a filesystem and you don't have to use it.

Nope I said three things:

-this is A MODULE and any code inside it does NOT affect anything else in the kernel

-there are code quality standards and this filesystem is apparently good enough (not that it is terribly hard, given how tiny and simple it really is), while Realtek's crap code for supporting their USB Wifi dongles (among other things) aren't even accepted in fucking staging.

-you have no clue and keep talking of nonsense politics


But feel free to strawman me