LWN: Comments on "linux-next on github"

linux-next on github

dmk — Fri, 23 Sep 2011 07:43:08 +0000

Their was one tweet teaser from the excited admin that it would be great and huge what was coming.

Objections to git?

bronson — Fri, 23 Sep 2011 02:00:32 +0000

I love git! It's fundamentally truly awesome with a few minor UI glitches. That's why I hope (tongue in cheek) that Linus rewrites Google+ in the same vein.

Maybe "G+" read as Git?

Objections to git?

mathstuf — Thu, 22 Sep 2011 20:25:57 +0000

> and I just can't get myself to use a GUI with it except for history browsing (gitk).

Have you tried tig[1]? ;)

[1]http://jonas.nitro.dk/tig/

Objections to git?

jrn — Thu, 22 Sep 2011 20:06:11 +0000

> What do you object to in (current!) git?

That it's a version control system instead of a Facebook alternative with better privacy features.

Objections to git?

vonbrand — Thu, 22 Sep 2011 16:24:41 +0000

What do you object to in (current!) git?

The above is not meant to be a tease, it is genuine curiosity. I first used RCS for a smallish project of mine (a half dozen files of code and some two dozen specification files to be processed by said code), and found it adequate for my limited needs but lacking. Then I tried to wrap my head around CVS and SVN, and failed. Next came BitKeeper, which I made friends with, and now I couldn't live without git; I used it to follow upstreams in SVN (but all of them moved on to git since, fortunately). I find git's model quite intuitive (but maybe my brain is wired all wrong...), and I just can't get myself to use a GUI with it except for history browsing (gitk).

linux-next on github

dlang — Wed, 21 Sep 2011 18:44:35 +0000

one thing to remember is that if they have called in the FBI to investigate the hacks, the FBI imposes a lot of gag restrictions.

I really wish we could get more information about what sort of progress that they are making. Based on what they've published so far I am taking their silence as indications that they haven't found anything else tinkered with (with the knowledge that the FBI may be trying to sit on them)

They are doing a lot of checking, and when you are talking about disk arrays as large as what they have (66 drives on a single machine), just the time to _read_ all that data in to the system (let alone getting the upstream version to compare to) can take several days.

linux-next on github

dlang — Wed, 21 Sep 2011 18:34:31 +0000

the fact that kernel.org takes the time to explicitly acknowledge the donation of individual servers (however large they are) is a good indication of how small they are.

linux-next on github

gcarothers — Wed, 21 Sep 2011 17:37:42 +0000

All of the source code for making the wiki is public

https://github.com/github/gollum

https://github.com/github/markup

linux-next on github

corbet — Wed, 21 Sep 2011 13:22:41 +0000

Actually, kernel.org has a full-time admin paid by the Linux Foundation. I suspect he's working rather more than full time at the moment, though.

linux-next on github

ebirdie — Wed, 21 Sep 2011 12:16:18 +0000

I think this isn't the whole picture of the problem.

Yes, certainly, this big and important service needs on-demand admin staffing and financial backing. Both are required to gain faster recovery, if the situation really is what is said above.

The situation (lack of financial backing, lack of full-time on-demand admins, raising voices "how the service can be down this long", service downtime having effects beyond convenience matters) are a proof those are lacking and needed, if the service is kept going on as before.

I think the whole picture also includes the problem of a centralized service and its size going up. Some time before the intrusion I read about news how the kernel.org service has got new hardware and old repurposed. All I was left to think about was, how big and complicated the service has become and the persons doing the admin work.

On one part I do feel desire to take a part on the challenge to keep up a service of size like kernel.org on volunteer basis, but on the other part I don't want it anymore. I have been on that grill for full payment and not sure, if it should be ok for admins to scale up. It wasn't ok for Linus to scale up in kernel development back in 10 years ago.

I see there is a technical and service modeling challenge to distribute the service and data under a service like kernel.org. One interesting technology to riddle oneself's design brain cells was lately introduced at lwn.net: Is it FileTea time?
https://lwn.net/Articles/458537/

The current mirroring model has pretty limited function today and, in respect to the above hint in FileTea, I think it pretty much wastes resources to what it seems worth in a service interuption case like this as the ultimate goal is to keep the service unaffected whatever happened. The mirrors do have their value, that is undeniable.

linux-next on github

andresfreund — Wed, 21 Sep 2011 08:05:46 +0000

I think some of the tardiness comes from the fact that there is about no information about whats going on out there.
I do realize that there is stuff that they won't publicize but giving some hint of progress/whats going on would go a long way.

linux-next on github

neilbrown — Wed, 21 Sep 2011 06:58:54 +0000

How many full-time admins do you suppose kernel.org has to work on this?

I'm not certain, but I think it is about 0.2.

I agree that it seems like a long time for it to be out, but I'll leave complaints about tardiness to the people who pay the bills and salaries for running kernel.org...

linux-next on github

rsidd — Wed, 21 Sep 2011 06:39:11 +0000

Really, what's up with kernel.org? Isn't it a bit embarrassing that, about 10 years after Linux declared itself "enterprise-ready", the key linux development server has been knocked offline for weeks thanks to an intrusion? If it had been, say, sourceforge or github that took itself offline for such an extended period, I wonder what the reaction would have been.

linux-next on github

bronson — Tue, 20 Sep 2011 17:10:27 +0000

If it inspires Andrew Tridgell to violate Google+'s questionable terms of contract ("Google has determined that the name 'Tridge' does not comply with proper naming standards"), forcing Linus to write a G+ that is actually worth a damn, then please please oh please let this happen!

linux-next on github

tzafrir — Tue, 20 Sep 2011 12:15:59 +0000

What would it take to use that in git.kernel.org or in your own personal git server?

linux-next on github

slashdot — Tue, 20 Sep 2011 09:53:08 +0000

Actually, I'd suggest github just bribed the kernel.org administrators to prolong the maintenance downtime beyond any reasonable time span.

linux-next on github

error27 — Tue, 20 Sep 2011 08:58:34 +0000

Yes and no. The move to git and linux-next is like the move from normal carriages to horseless carriages.

linux-next on github

neilbrown — Tue, 20 Sep 2011 08:32:25 +0000

<tongue location=cheek>
Base on the principle of "Cui Bono" and the fact that linux man pages just moved to github(*), I'm going to suggest that it was a github initiative to take kernel.org down.
</tongue>

(*) http://linux-man-pages.blogspot.com/2011/09/relocated-git...

linux-next on github

geertj — Tue, 20 Sep 2011 08:02:19 +0000

Actually the github wiki offers a set of standard markup formats (Textile, RestructuredText, Markdown, ...), and its contents are available as a git repo. So i wouldn't call it proprietary either.

linux-next on github

tzafrir — Tue, 20 Sep 2011 05:51:58 +0000

That depends: If you use it merely as a git repository, it is as good as any git hosting.

If you start to depend on non-standard features of it (e.g.: bug reports, wiki), you can call it a proprietary infrastructure, is switching from it won't be trivial.

linux-next on github

tzafrir — Tue, 20 Sep 2011 05:49:43 +0000

I very much hope not. I wonder what will be the git of Google Plus.

linux-next on github

martinfick — Tue, 20 Sep 2011 03:55:25 +0000

I suspect that it actually means that many of the lieutenants are getting an opportunity to code for Linux themselves a little more, instead of mostly reviewing and merging patches. They might be able to quietly tackle some of the harder stuff that always gets put off, maybe even some of the stuff that only they can tackle!

linux-next on github

Cyberax — Tue, 20 Sep 2011 01:33:10 +0000

-mm tree was the de-facto linux-next back then.

linux-next on github

Lennie — Mon, 19 Sep 2011 23:20:29 +0000

Well, it is not like Bitkeeper where the community did depend on Bitkeeper (for example the application).

With git this is not true, as has been shown now.

It is actually pretty easy to switch the distribution point to somewhere else.

linux-next on github

elanthis — Mon, 19 Sep 2011 22:30:04 +0000

I'm honestly surprised nobody is bitching about using a proprietary service/infrastructure. :)

(note: I <3 github)

linux-next on github

error27 — Mon, 19 Sep 2011 22:10:22 +0000

It's good that linux-next is up again. :) It's amazing how we used to operate without it back in the day.

linux-next on github

proski — Mon, 19 Sep 2011 22:02:43 +0000

Subscribing to our esteemed editor's page in case it happens. It's very frustrating to have Linux development disrupted for so long. The only upside I can think of is that kernel hackers are bringing their expertise to non-kernel projects in the meantime, such as their favorite development tools.

linux-next on github

Cyberax — Mon, 19 Sep 2011 17:00:22 +0000

So, distributed version control systems really do work? :)

It's nice to see that the kernel community can work around infrastructure problems. What's next, Google+ in case LKML servers fail?