Why isn't F2FS support in the Kernel?

Gerald B. Cox gbcox at bzb.us
Mon Dec 22 20:41:29 UTC 2014


Please accept my apologies.  My initial post was sufficient to make my
point.


On Mon, Dec 22, 2014 at 12:29 PM, Stephen Gallagher <sgallagh at redhat.com>
wrote:

>
>
>
> On Mon, 2014-12-22 at 11:57 -0800, Gerald B. Cox wrote:
> >
> > On Mon, Dec 22, 2014 at 11:31 AM, Stephen John Smoogen
> > <smooge at gmail.com> wrote:
> >         No they do not have all the information needed. What they know
> >         is that some other distribution ships it and that it works in
> >         a device using a custom kernel. How does it work on a normal
> >         drive, how does it not work, how are the tools functioning
> >         with the toolset, what extra patchsets need to be found and
> >         gotten to make sure what is in the kernel actually works as
> >         well as it does in say the Nexus 9.
> >
> > Oy Vey!  This isn't a space shuttle launch.  If no one else was using
> > this, that would be another thing.  You're also making up rules that
> > weren't applied to other products which are included in Fedora; and
> > asking for Q/A theater to obfuscate.   You can try to spin it another
> > way, but most people aren't buying it.
>
> Gerald, please moderate your tone. You are discouraging people from
> listening to you.
>
> To be clear, I'd like to use a car metaphor (because that's what we do,
> right?) to help you recognize your behavior, that you might learn from
> it and be more helpful in the future:
>
> You drive your brand-new minivan up to the local racetrack. You hop out
> of the driver's seat and walk up to the nearest mechanic. You say to
> this mechanic "Hey, I just heard that over in Raceville they have a guy
> that put a Hemi in a Sienna. Stick one in my van over there and I'll
> race it." The mechanic stares at you, confused. He says to you "I don't
> have any experience performing that sort of operation, nor do I have the
> tools. And it's not a set of skills I can see being widely useful, so
> it's not worth my time to learn how and buy the equipment to do it."
> Meanwhile, you get angry and complain that "Well, the other guys can do
> it, so you must be able to do it too!"
>
> (Of course, in that metaphor, I'm assuming you're *at least* going to
> offer to pay the mechanic to do the work. When you came in here and made
> your demands, it was strongly implied that you expected someone to
> expend their own time and money to please you, which is also not a good
> way to encourage people to do what you want.)
>
>
>
>
> Now, you are misunderstanding the level of effort necessary to get
> certain features into Fedora. It's comparatively easy to get a new
> application added to the distribution because it's self-contained. If it
> doesn't work, it will have no impact outside of itself.
>
> Traveling further down the stack, you start getting into packages that
> are required dependencies for other packages (such as Django or Rails).
> These require at least an order of magnitude more care and feeding
> because of the number of other packages that depend on them. It takes a
> more committed individual to include that in the distribution.
>
> Now let's go a little further down the stack to the platform layer. Now
> we have things like the python platform and glibc. These are packages
> that are depended on by thousands of other packages. Maintaining any one
> of these is likely to be the full-time job of at least one person (and
> likely a whole team of them).
>
> Now let's go even further down to the kernel (and specifically, the
> filesystem layer). We are now at pretty much the absolute lowest level.
> Everything on your installed system depends on this working and with no
> critical issues. This is the full-time job of dozens of people, with
> specialists in certain particular drivers. The filesystem layer is
> extremely fundamental, as bugs in that layer usually mean that data is
> lost or performance is unacceptable. This causes far-reaching issues.
> That's the reason that everything that goes into the kernel is *very*
> carefully vetted and tested. Well, most of the time; to use the specific
> example you cited earlier in this thread, btrfs went in far before it
> was ready because the btrfs developers committed to dealing with the
> fallout. The btrfs developers have repeatedly and publicly stated that
> btrfs is not production-ready (regardless of what certain other distros
> claim) and Fedora is wise to listen.
>
> F2FS is perfectly welcome in Fedora, as long as a sufficient set of
> people are willing to do the stabilization and testing work necessary
> for that inclusion. Demanding that a feature you want must be in the
> distribution is not only unhelpful, it's actually insulting to all of
> the people who work hard to see that Fedora is both leading *and*
> actually stable for use. If you are not capable of maintaining it, then
> your best bet would be to go to the *upstream* developers of F2FS and
> ask *them* to volunteer to maintain the driver in Fedora. That would
> have a far higher chance of success than ranting on the Fedora lists.
>
>
>
> --
> devel mailing list
> devel at lists.fedoraproject.org
> https://admin.fedoraproject.org/mailman/listinfo/devel
> Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.fedoraproject.org/pipermail/devel/attachments/20141222/71d548f4/attachment.html>


More information about the devel mailing list