| |||||||||||||
Who wrote itWho wrote itPosted Apr 1, 2008 15:10 UTC (Tue) by coriordan (guest, #7544)Parent article: The Linux Foundation's kernel development paper
I think many people will take a misleading view from this about who free software is written by. Take the hypothetical case of the guy who wrote the widget allocator in the early 90s. It was tweaked in 95 and 98 to be more stable, and largely replaced in 2000 with a faster version. In 2004, it was rewritten to be fairer etc. Looking at it in 2008, one concludes the widget allocator code was written in 2004 (probably by some huge corporation). But *the work of developing a widget allocator* was not all done in 2004. The 2004 work was just writing new code based on a working implementation (that the rewriter calls brain damaged, fubar, etc.) and a decade of feedback. This document accurately documents who is currently writing kernel code, but I think it would be really interesting to see who *wrote* the kernel. Of course, that's harder research to do. I haven't read this whole document, but if there's no caveat about misinterpretting the findings, it might be worth adding one. (I already posted this comment, but my formatting was poor so I think an LWN filter rightly discarded it.)
(Log in to post comments)
OT: filters Posted Apr 1, 2008 15:20 UTC (Tue) by corbet (editor, #1) [Link] (I already posted this comment, but my formatting was poor so I think an LWN filter rightly discarded it.) We don't have any filters - automated or otherwise - discarding comments. If you truly posted a comment - as opposed, say, to stopping at the "preview" screen - and it didn't appear, we want to know about it. That should not happen.
OT: filters Posted Apr 1, 2008 16:43 UTC (Tue) by coriordan (guest, #7544) [Link] Aha. State "P" (on my Comments posted page) must mean I clicked the button to preview it but forgot to then approve it. Ok, thanks for clearing that up. My mistake.
Who wrote it Posted Apr 1, 2008 16:34 UTC (Tue) by AlexHudson (subscriber, #41828) [Link] But equally, you can't compare new code and old code by what it says it does on the tin. If the 2004 version was totally rewritten, that could indicate serious problems with the older code that couldn't be remedied in an evolutionary manner: even if the old code allocated widgets, if it did so in such a manner that it was basically unusable to some large proportion of people, then it's not really relevant that the "feature" was around in 2000 or earlier.
Who wrote it Posted Apr 1, 2008 16:57 UTC (Tue) by coriordan (guest, #7544) [Link] Even if a rewrite is truly needed, the work of the first programmer is still more than zero and the work of the rewriter is still less than 100% of the task. I'd guess in most situations, the rewriter did less work than the guy who wrote the original. My point being that the first 6 years of development was more important in terms of moving the project from non-existant to perfect than the work in the subsequent 12 years, but the analysis in the linked document attributes importance in the opposite way. Imagine a company who spent the nineties trying to keep free software back by denying us documenation, but who then hired a kernel hacking team in the last five years to sqeeze out 5% better performance here, 10% there, and scale it up to extreme case. That company "scores" highly in the linked document, while hackers who did great work when we had little, like Alan Cox, "score" quite low. There's nothing false in the linked document, I just think a lot of people will misinterpret it.
Who wrote it Posted Apr 1, 2008 16:59 UTC (Tue) by zlynx (subscriber, #2285) [Link] Even if the "widget allocator" had horrible problems, the original author still merits credit. Why? Because, like Edison said about the lightbulb, he discovered some of the many ways *not* to do it. That is a very useful discovery.
Who wrote it Posted Apr 1, 2008 17:22 UTC (Tue) by gregkh (subscriber, #8) [Link] > I think many people will take a misleading view from this about who free > software is written by. That was not the goal of this paper. The goal was to report who had been doing the work in the past 3 years. Your goal is a nice one, and can easily be done if you dig through the commit logs, but again, was not the goal of this paper, so that is why it is not covered.
Who wrote it Posted Apr 1, 2008 19:01 UTC (Tue) by coriordan (guest, #7544) [Link] It's just a misinterpretation/misuse that I expect. I've seen people cite a similar article posted one year ago (Who wrote 2.6.20?) as evidence that GNU/Linux is a corporate project rather than a community one.
Who wrote it Posted Apr 1, 2008 18:53 UTC (Tue) by beoba (guest, #16942) [Link] When kernel development was moved to git, was commit history preserved in some way? If so, does git have an equivalent to the "blame" command? If so, you could get some interesting statistics describing who (both people and companies) have had the most influence in which components. I've only had much experience with centralized SCM's, so I don't know whether this sort of thing is actually possible with git/bitkeeper.
Who wrote it Posted Apr 1, 2008 18:58 UTC (Tue) by beoba (guest, #16942) [Link] After having read past the first few paragraphs, I noticed that it already has some charts for this :)
Who wrote it Posted Apr 1, 2008 19:08 UTC (Tue) by nix (subscriber, #2304) [Link] The git equivalent for the blame command is... blame! :)
|
|||||||||||||