Gentoo Developers Unhappy, Fork udev

[ryao]

They force dependencies on you all the time, just like all other projects. In this case, it happens to be a dependency that you disagree with.
Yes, I know it "would make GNOME Linux only", but there are some points to consider:
1) Isn't that the de facto case already?
2) The dependency is IIRC on the systemd interface, meaning that you can replace it with other software that provides the same functionality.

1. GNOME replaced CDE on Solaris as the only provided desktop environment not that long ago. It runs on more than just Linux.
2. That gives distribution developers more work.

Who are 'we'? Are you now speaking for Gentoo as a whole, I really hope not because I doubt I could fathom a worse spokesperson.

And while I'm grateful for the work Canonical has put into making Linux a 'mainstream' desktop possibility, there's no comparing Red Hat's and Canonical's contributions to the Linux ecosystem, Oracle doesn't even register.

It is a hypothetical question to which I gave a hypothetical answer. It was left vague intentionally. Also, Gentoo doesn't have a spokesperson. Each Gentoo developer's opinions are their own.

As for RedHat's contributions, they tend to focus on quantity over quality. With the exception of Linus' tree, patches that they provide to upstream projects often lack the scrutiny given to patches that they write against the Linux kernel. This seems to be in part because they only seem to care about their narrow set of target cases and in part because they do not seem to enforce QA standards on what they submit. It is nice of them to try to contribute, but there are plenty of people who would prefer RedHat employees to enforce better QA standards on their contributions and to handle cases that they do not necessarily need themselves. It would prevent regressions that such patches often cause for others.

By the way, I finished reworking the kmod builtin in eudev. It has been committed to HEAD.

[kigurai]

1. GNOME replaced CDE on Solaris as the only provided desktop environment not that long ago. It runs on more than just Linux.

Yes, I know it runs on more than Linux. The point was that all development seems to be done by Linux people. That is the impression I had when the systemd-in-Gnome issue surfaced the first time around.
Also, the WP article on JDS/OpenSolaris says that "OpenSolaris Desktop is now tied to the OpenSolaris operating system", which makes your point kind of moot

2. That gives distribution developers more work.

That's what usually happens when you decide to go against the stream. But I guess you know that already, with the fork and all

[funkSTAR]

As for RedHat's contributions, they tend to focus on quantity over quality. With the exception of Linus' tree, patches that they provide to upstream projects often lack the scrutiny given to patches that they write against the Linux kernel.

What a load of crap. Linus' tree doesnt cover any consumer level QA. He is tagging a kernel for stable release the minute he feels the patch level has calmed enough. Reason: Waiting longer creates a larger thus more challenging patchset for the next kernel. This is really awful. Mainline QA is a regression ride as well, so much crap enters stable releases as well. This means distros gets a nonstable kernel with a stream of regressions from stable updates all the time. RH is willing to eat this shit and provide a first frontier for QAing linux. It is called Fedora.

And I think RH does a wonderfull job with Fedora. Without this QA effort everything would suck ten times more. And for claims about RH not caring for quality? Go ask the paying customers.

[ryao]

What a load of crap. Linus' tree doesnt cover any consumer level QA. He is tagging a kernel for stable release the minute he feels the patch level has calmed enough. Reason: Waiting longer creates a larger thus more challenging patchset for the next kernel. This is really awful. Mainline QA is a regression ride as well, so much crap enters stable releases as well. This means distros gets a nonstable kernel with a stream of regressions from stable updates all the time. RH is willing to eat this shit and provide a first frontier for QAing linux. It is called Fedora.

And I think RH does a wonderfull job with Fedora. Without this QA effort everything would suck ten times more. And for claims about RH not caring for quality? Go ask the paying customers.

Linus' tree requires proper commit messages and a fair amount of discipline in what can constitute an individual commit. That alone goes a long way toward better maintainability.

[funkSTAR]

Linus' tree requires proper commit messages and a fair amount of discipline in what can constitute an individual commit. That alone goes a long way toward better maintainability.

No. Formatting and commit messaging is _trackability_ not quality. Of course it also serves as a sanity filter; If you are unable to adapt your patch then you are bat shit crazy.

Code quality can happen by accident even without trackabillity. QA is continous validation on said code. Fedora(RH) does that, Linus doesnt.

[jntesteves]

Let me try to summarize the problem here:

1. Udev is broken since forever. Nobody seems to care. Nobody is willing to maintain or contribute to udev;
2. Udev is merged to systemd;
3. Udev problems are fixed;
4. A couple broken drivers written for a broken subsystem fail and need to be fixed too;
5. Some people don't agree with the fixes in udev and fork it to fix it their way;

So everything is fine. Just another day in an open-source environment. Copyrights where violated, people trying to offend each other through public discussion means. But that's just the community part of it, so, who cares? I care more about working code written.

[alcalde]

I still find it funny that people choose to ignore what linux tells them. It's like they assume he's an idiot and go on do their thing. If torvalds says what you do is shit maybe you should at least consider him.

Linus <> Linux. He hates everything and everything that isn't exactly the way he would do it is garbage in his mind.. If we considered every negative comment from Linus we'd have to throw everything out except the kernel he doesn't code for anymore. It's just like if we listened to Stallman we'd all be accessing the web through wget and shunning wifi. Like every other open source contributor, they have their strengths and serve their purposes, but the Linux platform doesn't depend on, or take orders from, any one of them.

[a user]

And I think RH does a wonderfull job with Fedora.

i don't think so.

And for claims about RH not caring for quality? Go ask the paying customers.

paying customers usually pay for blending features and ignore quality issues. best example of all is the big win of windows. so your suggestion is the worst possible idea for proving that a software company is caring for quality.

[a user]

Linus <> Linux. He hates everything and everything that isn't exactly the way he would do it is garbage in his mind.

seems his statements were to compex for you to undestand. i know it is easier to interpret them that way for a simple mind.