Weekly Edition
Return to the Announcements page
| |||||||||||||
Linux Foundation kernel authorship paper updated
The Linux Foundation has announced
the release of an
updated study of kernel code authorship [PDF] written by Greg
Kroah-Hartman, Amanda McPherson, and Jonathan Corbet. "The updated
study finds that since April 2008, there has been a 10 percent increase in
the number of developers contributing to each kernel release and that a net
of 2.7 million lines of code have been added. This level of activity has
resulted in an average of 5.45 patches being accepted per hour, an increase
of 42 percent since the original study." The information will be
familiar to LWN readers, naturally, but it's a slicker presentation and
it considers a longer time period than LWN usually does.
(Log in to post comments)
Linux Foundation kernel authorship paper updated Posted Aug 19, 2009 17:50 UTC (Wed) by patrick_g (subscriber, #44470) [Link]
In the pdf properties, for the creation tag there is "Adobe InDesign CS4 (6.0)".
This software doesn't run on Linux and I think it's a shame to see this sort of thing from the official Linux foundation.
Linux Foundation kernel authorship paper updated Posted Aug 20, 2009 15:16 UTC (Thu) by kragil (subscriber, #34373) [Link]
Well, most PDFs in the Linux world are done with InDesign (Ubuntu dev week brochure is another example, but you will find more.) .. that is the sad reality of things.. I guess the design monkeys are not able to learn good programms like Scribus.
Linux Foundation kernel authorship paper updated Posted Aug 20, 2009 22:16 UTC (Thu) by MattPerry (guest, #46341) [Link]
> I guess the design monkeys are not able to learn good programms like
> Scribus.
More likely it's that the design firms that this work is outsourced to use InDesign because it's largely becoming the industry standard for publishing layout. Also, there is no equivalent for Acrobat Pro or Distiller on free software operating systems, so creating print-ready PDFs with all the effort and attention to detail that such action requires cannot yet be done on a free OS. Of course, there are still many commercial printers that would prefer to receive the original documents in InDesign or Quark format rather than deal with PDFs.
Counting the wrong thing Posted Aug 19, 2009 18:53 UTC (Wed) by coriordan (guest, #7544) [Link]
I've just skimmed it, and it's full of great info, but it seems to make the same mistake previous similar efforts have made: it measures typing rather than development.
How about we apply the same method to a translation of Ulysses? We will find that >99% of the book was "written" by the translator, and somewhere between 0 and 1% was written by James Joyce.
It's the same with the Linux kernel. We had a working kernel in the mid-90s before any corporations were really contributing. The memory management has seen been rewritten, but does that mean the person who got us from zero to working counts for nothing and the person who squeezed out a 15% speed increase should get all the credit?
This paper says that the largest contributor to the kernel is people without an affiliation (24%), but I think that's undervaluing the work of the people who did the really essential work to get us from zero to working kernel, and giving all the praise to the people who did great, useful work to get us from working kernel to better working kernel.
What this study shows is who *maintains* the Linux kernel.
Linux Foundation kernel authorship paper updated Posted Aug 19, 2009 20:35 UTC (Wed) by davem (subscriber, #4154) [Link]
Well, who do you think is in the group of people who got us from "zero to working" and who is in the group who got us "working to great"?
A lot of people are in both groups.
And keep in mind that people who are able to review and integrate thousands of patches every year into the kernel must have a serious understanding of the code they care for. That doesn't come overnight and, surprise surprise, a lot of that comes from having written the original code in the first place.
Linux Foundation kernel authorship paper updated Posted Aug 19, 2009 21:45 UTC (Wed) by coriordan (guest, #7544) [Link]
I think the people that got us from zero to working is a group of 100 to 500 developers who were active from 1991 to 1997. The study says there are 4,910 developers active today. No research needed to know that most of the 4,910 can't have been in the group of 100 or 500 :-)
Reading this sort of count, one would think that we have IBM+Novell+Intel to thank for 18.4% of the kernel.
I see a lot of late joiners on that list, some of whom spend the 90s trying to make life difficult for the Linux hackers of the day, and now they waltz in with their spanners, do an admittedly great job of tightening the nuts, but then get credit for building the thing.
I think the hackers who were there in the 90s (I might be talking to one) get too little credit, and big firms with big PR departments are getting too much.
Linux Foundation kernel authorship paper updated Posted Aug 20, 2009 7:35 UTC (Thu) by mingo (subscriber, #31122) [Link] A couple of comments. (and like DaveM i've been around in '97 too) Firstly, i think it's a given that any 10-page paper that tries to summarize almost two decades of kernel development done by thousands of people who authored more than a hundred thousand commits resulting in more than 11 million lines of code with 1 millions of lines of code flux every 3 months is going to be incomplete, inevitably. Secondly, while it took fewer people to bring Linux from 'nothing to working' initially, it takes a lot of resources to keep the thing still great and current. [ I hasten to add that after we reached the "90% ready" status in 1997 it also took a lot of resources to complete the remaining 90% of the work ;-) ] If anyone took 1997's 'working' kernel and tried it on today's hardware he'd quickly find that it 1) possibly wont boot at all 2) wont recognize your SATA drive 3) wont handle your USB keyboard 4) wont scale very well on your quad 5) wont read your flash cards 6) wont detect your ethernet card 7) wont detect more than 1GB of RAM 8) wont filter incoming packets ... and the list of show-stoppers goes on. So the kernel we were hacking proudly in 1997 we wouldn't be all that proud of today - weren't it for those thousands of "new" people who came on board.
Linux Foundation kernel authorship paper updated Posted Aug 20, 2009 7:53 UTC (Thu) by hppnq (guest, #14462) [Link] Also, I think that releasing drivers for your own hardware counts more as sponsoring your own company than sponsoring kernel development -- I won't even mention sponsoring Free Software development. In the paper it states that this benefits the user, but that is not at all the point of Free Software, as RMS pointed out so many years ago.This paper looks at Linux kernel development exclusively from the company point of view -- and there is really, really nothing new there. So there are four ways of stating that Linux development is profitable, that's great.
markets didn't write the Linux kernel Posted Aug 21, 2009 1:24 UTC (Fri) by coriordan (guest, #7544) [Link]
Yeh, I think that's an important point about the difference between supporting your company and supporting Linux kernel development.
I guess the reason this method of counting irritates me is that it makes it look like a bunch of market-driven profit-orientated companies are the authors of the Linux kernel. A reader would think that market forces causes the Linux kernel to exist, but it wasn't that, it was people's love of tinkering and people wanting to have a code-sharing community kernel, and people wanting features that the market wasn't providing. The market forces only kicked in after the zero-to-working work was done (much of which was unpaid).
markets didn't write the Linux kernel Posted Aug 21, 2009 13:21 UTC (Fri) by mcmanus (subscriber, #4569) [Link]
"The market forces only kicked in after the zero-to-working work was done"
That's the GPL at work, right? Once the product has reached critical mass, the 3rd party gets more by using it than it pays in competitive advantage by releasing its own addition in GPL comlpiance. Before some inflection point that trade-off isn't true (or at least, isn't obviously true).
As Linux gets ever more feature rich this is a stronger advantage for it each day.
markets didn't write the Linux kernel Posted Aug 21, 2009 13:52 UTC (Fri) by coriordan (guest, #7544) [Link]
Exactly. When there's a copyleft licence to ensure that everyone will have to play fair, there comes a point in a free software project when its maintenance and continued development can be mostly left to corporations. The Linux kernel reached that point years ago.
But the corporations can't be relied on to do the essential work before that point. For that we need people who value freedom, people who love tinkering, groups who like writing and sharing software, etc. I think this group does the more important job but is often forgotten.
|
|||||||||||||