LWN: Comments on "Developer statistics for 2.6.30"

Canonical? No.

snadrus — 2009-08-12T19:22:07+00:00

Important point!
If [their organization cannot persist (which is true if they live in the red)] then
anyone requesting more from them could at-best see short-term benefits until they run out of cash reserves.

They've enhanced the open source desktop with Upstart & invited more independent developers to open source with Quickly and Launchpad.

Canonical?

jmm82 — 2009-07-21T03:34:38+00:00

I have been using Ubuntu for two years, exclusively. Before Ubuntu I used Red Hat.

I work at a start-up with about 10 Linux users and all but one use Ubuntu(the other uses Suse). Everyone I know that uses it are happy and yes we do *real* Linux programming stuff.

Canonical? No.

job — 2009-06-29T20:17:14+00:00

You mean they lose money on it. :)

I will save any judgement on Canonical or any other company until they have black figures. Until then, the process is not yet sustainable and matters little in the long run.

Only 1/3rd!

jengelh — 2009-06-27T12:54:38+00:00

There is no need to do initial commits with a separate user id. I just need to adjust the script

Total lines: 11934488
  #1     4246304  35.58%  *initial checkin
  #2      230061   1.93%  Greg Kroah-Hartman
  #5       85543   0.72%  David S. Miller
#178       10621   0.09%  Linus Torvalds
#229        8424   0.07%  Andrew Morton

Canonical? No.

mjg59 — 2009-06-22T22:32:21+00:00

It's entirely untrue that Canonical contribute nothing to upstream. It can certainly be argued that their contributions are small relative to their profile and developer count, but that's not the same thing.

Canonical? No.

asdlfiui788b — 2009-06-22T20:14:45+00:00

Canonical does not contribute upstream to Linux, at all. Whatsoever. They just take what they can scrap up from Debian, and make money on it.

scripts available?

jake — 2009-06-11T18:10:54+00:00

> Are the scripts/tools used to generate these stats publicly available?

yup, the repository is here: git://git.lwn.net/gitdm.git

This message might be of interest as well: http://lkml.org/lkml/2009/1/14/452

jake

scripts available?

khilman — 2009-06-11T15:46:40+00:00

Are the scripts/tools used to generate these stats publicly available?

Canonical?

jengelh — 2009-06-10T19:25:35+00:00

Well handholding seems scarce if I have to look at some Launchpad projects, so there is not much left.

Canonical?

SimonO — 2009-06-10T07:18:43+00:00

Perhaps it doesn't matter too much either way? Kernel contributors tend to be individuals rather than companies, so if one company would have to lay off people, other linux companies who have more success can hire them.

It would be worrying if the company which is more successful would not allow upstream contributions.

/Simon

Canonical?

Duncan — 2009-06-05T02:22:18+00:00

> On the other hand, I have met many people
> who have a Ubuntu Live CD, or a partition,
> which they boot probably once a month. This
> is a fake user base. Their primary OS with
> very few exceptions is Windows.

Agreed in general that these aren't true users, and that there are
probably a lot of Ubuntu LiveCDs and installs in that group. However,
it's important to note that the *ix way of computing is a decently large
change in computing mindset, and that some portion of these will
ultimately become full-time Linux users.

(FWIW, I was in this segment for 2-3 years, around the turn of the
century, with Mandrake the distribution I was playing with. Then MS
decided they were going a different way than I was, and basically gave me
that last push I needed to make the jump, when they went the eXPrivacy and
remote authorization route. This for a user who had previously considered
most MS software "too important" to risk warezing, and who had previously
been spending probably 50% of his computing dollars on MS directly. But
after the push, I soon discovered how liberating the land of freedomware
was, and now look back at proprietaryware much like a defector looking
back at his former home -- I have a lot of friends and family I left
behind and will do what I can to help them make the jump as well, but
it's nowhere I want to be or can even visit, unless the regime changes and
becomes free as well. I no longer even consider proprietaryware, nor
could I without serious legal issues, as I can no longer agree to all
the !#@! demanded of proprietaryware users, whose masters, those supplying
the proprietaryware, really /do/ seem to think of them as slaves, not
actual human beings, with few if any rights worth considering.)

> These are probably the people who upgrade
> their Ubuntu every 6 months - I am sure
> that nobody who actually uses their computer
> would be crazy enough to do an upgrade so
> frequently.

Why not? I actually use my computer, running Gentoo, and upgrade on
average 2-3 times a week. In fact, I hate going a full week without an
upgrade as the changes start getting too large to easily cope with all at
once. Thus, I'd argue that six month upgrades, far from being too
frequent, are WAYYY too infrequent, by three orders of magnitude!

FWIW, I found the same issue on Mandrake and quickly switched to Cooker,
where the rolling updates were MUCH easier to cope with and most of the
big issues the release version upgraders had to cope with were incremental
changes I had dealt with as a matter of course tiny incremental bits at a
time, long months before. Six months is simply WAY too long for easy
upgrades, and going longer than that, you're needlessly losing out on
updates and features that make computing both easier and more pleasant.

Duncan

Canonical?

Lennie — 2009-06-04T07:29:43+00:00

Well, it all depends how you look at it, if webstatistics are anything to go by, Linux has something like 1% of the desktop. That means 1 in a 100 is a Linux-desktop. Have you seen what a 100 desktops are running in the past months ? I know I haven't. Actually, possible not even in years (ok, if I don't count the LUG-meeting I visited). I also know their are a lot of people who are like minded and they use Linux 'together'. So these Linux desktops also are clustered together. That means it's even less likely you'll see a Linux desktop in the wild. I use it every day and I know 5 people who use it atleast a few times a week.

Canonical?

dlang — 2009-06-02T21:33:51+00:00

I know a couple dozen people who use ubuntu as their distro, all day long.

I think it is going to depend a lot on what groups you are with as far as how many people use what distro.

I used slakware for 10+ years so I'm not a non-technical user, and I decided to switch to ubuntu.

Canonical?

mikov — 2009-06-02T18:24:55+00:00

Yeah, I have lots of "evidence"; that is what I do all day - gather evidence to implicate Canonical in trying kill Linux :-)

Seriously, I think these are valid questions to ask and discuss if we do it politely. The consensus between those who responded is that Canonical is benefiting the Linux ecosystem and I agree.

I do however think think that there are two different alternate realities here. One is the overenthusiastic reality which you see at tech sites and discussion forums (e.g. Slashdot), and the other is the actual physical reality.

For example In the physical reality I have never met a person who actually uses Ubuntu. When I say "uses", I don't mean install every new version and "try" it, but actually use it for everything on his home/work PC every day. I am sure such people exist, but they are much fewer in absolute numbers than one would assume by reading tech sites. (I myself actually _do use_ Ubuntu on my laptop 100% of the time, but people think I am crazy. We also tried to use it on desktops at work, but alas that failed).

On the other hand, I have met many people who have a Ubuntu Live CD, or a partition, which they boot probably once a month. This is a fake user base. Their primary OS with very few exceptions is Windows. These are probably the people who upgrade their Ubuntu every 6 months - I am sure that nobody who actually uses their computer would be crazy enough to do an upgrade so frequently.

So, I think the answers are more complicated.

Canonical?

nix — 2009-06-02T12:55:11+00:00

Even if it was a zero-sum game, the kernel works quite well and gets
vastly more development attention than any other component of the system.
Perhaps it would be a good thing if other components (e.g. X) got lots of
new developers :)

Canonical?

martinfick — 2009-06-01T23:10:00+00:00

Do you have any evidence that they are taking away RH or Novell customers? I think it's more likely that all 3 are mostly growing new customers.

Canonical?

rahulsundaram — 2009-06-01T05:50:45+00:00

Support can either mean hand holding (ie) do foo to accomplish bar or it can be mean prioritized bug fixes or new features and if it is the latter and the vendor is pushing those fixes upstream, then it would show up in these statistics.

Canonical?

mikov — 2009-05-30T22:19:32+00:00

If they provide kernel support, wouldn't it result in patches, which then would have to visible in the above statistics?

(BTW, I want to reiterate that I have nothing against Canonical, and in fact at one time our company was considering purchasing support. Alas, instead there was a company-wide downgrade from Kubuntu to Windows XP for all non-developers. That however is a different subject...)

Canonical?

kragil — 2009-05-30T10:06:33+00:00

Canonical employs a few very capable kernel hackers. Sure they don't contribute as much as the RH or Novell guys, but maybe that it because their time is spent providing high quality kernel support. Think about it ;)

Canonical?

mikov — 2009-05-30T01:12:34+00:00

Thanks for the reply. I generally agree with what you are saying - more Linux adoption is a good thing, and of course there is neither a law, nor an ethical requirement, saying that everybody has to contribute to the kernel.

However there is a secondary observation that has been bugging me for some time: in light of these statistics, it seems less likely that Canonical would be able to provide high quality kernel support with their support licenses.

Canonical?

dlang — 2009-05-29T19:43:28+00:00

they don't do much kernel development, they focus on the userspace stuff.

remember that opensource isn't a zero-sum game. yes ubuntu takes some customers away from RedHat, but they also get a lot of people using linux who would not have used RedHat.

some small percentage of these additional people that they get involved with linux will become kernel developers someday, so even with no direct payments to kernel developers they still benifit the kernel

Canonical?

mikov — 2009-05-29T18:55:22+00:00

I hope I can ask this without sounding trollish - is Canonical present at all in these statistics? Do they do any kernel work?

Please note that I am not criticizing, just asking. Everybody is free to use whatever business model they can with Linux.

I have to admit though that I am a little concerned that if Canonical is taking customers away from RedHat and Novell (both major kernel contributors), it could ultimately hurt the kernel.

Developer statistics for 2.6.30

iabervon — 2009-05-28T16:30:51+00:00

It might be interesting to look at the non-author sign-offs by person as a percentage of changes, rather than as a percentage of non-author sign-offs. It's not as interesting to know how much of the total gatekeeping is done by someone as to know how much of the total change is through them. In this case, 10.4% of the changesets got David Miller's sign-off in transit.

Only 1/3rd!

incase — 2009-05-28T16:29:49+00:00

That's one reason I usually do initial commits with a distinct user id on any repository.
That makes it easy to see which code was in the initial commit (and would therefor possibly need to be checked for ownership in an older repository).

The cleaner solution of course is to get the hostoric information from the old repository also into the new one, but that doesn't always work (as, probably, during the BitKeeper->git move).

Only 1/3rd!

nix — 2009-05-28T13:40:40+00:00

It could only do that reliably if he wasn't writing any code at all anymore.

Only 1/3rd!

alex — 2009-05-28T09:00:44+00:00

That's an interesting number. I wonder if the "blame" for Linus (or possibly
more correctly the initial commit) can be trended to show when the kernel
will have been fully re-written PG (Post GIT).