| |||||||||||||
|
| |||||||||||||
Is this true?Is this true?Posted May 3, 2005 14:37 UTC (Tue) by ccyoung (guest, #16340)Parent article: Linus Torvalds' BitKeeper blunder (InfoWorld) Tridgell protested. He hadn't violated any license, he said, because he'd never agreed to one. He'd never even used McVoy's BitKeeper client; he merely intercepted the server's communications as they went across the wire and decoded them.
(Log in to post comments)
Is this true? Posted May 3, 2005 14:56 UTC (Tue) by beoba (guest, #16942) [Link] ..well, sorta. Check the link. He didn't even have to sniff any packets, as that line implies.
Is this true? Posted May 3, 2005 15:55 UTC (Tue) by sbergman27 (guest, #10767) [Link] Yes. The "reverse engineering" was so trivial that it's hard to even call it that.
And I agree with the article. The end result of the Bitkeeper situation was so very predictable. This is one of those rare cases where Linus made a mistake and then refused to listen when people pointed out that it was a mistake. (And I'm usually a Linus Torvalds cheerleader and not a critic.) He made what would have been a very pragmatic decision in any closed source environment, and it turned out that it was the least pragmatic decision he could possibly have made in an open-source situation. Did anyone honestly believe that the issues could ever have been worked out?
Is this true? Posted May 3, 2005 16:25 UTC (Tue) by emkey (guest, #144) [Link] I'm sorry, I still don't see what Linus did as a mistake. Well, there has been some bad PR recently but frankly people will say silly things sometimes.
BitKeeper allowed the Linux kernel to make huge strides the past few years. When it was no longer a viable solution Linus changed directions. Thats all that happened here. It is a big mistake to make this into some sort of idiological test case because the evidence in support of that point of view is very weak.
Is this true? Posted May 3, 2005 16:42 UTC (Tue) by sbergman27 (guest, #10767) [Link] Well, if it was not a mistake and everything played out in such a controlled fashion, why is Linus so upset? As another reader pointed out, he felt the need to use the term "dammit" in discussing the situation eight times in this weeks "Git Traffic", which is rather uncharacteristic of him. Add that to his public flame at Tridge (for telneting to a Bitkeeper server and typing 'help', no less) and a picture emerges of someone who feels they just bludered publicly. The kernel project may have gained some practical benefits from using Bitkeeper, but I believe in calling a spade a spade. If this work on git had been started 3 years ago, we'd be in better shape today.
Is this true? Posted May 3, 2005 16:52 UTC (Tue) by emkey (guest, #144) [Link] My guess is that he's pissed because a lot of immature people are giving him grief. Without Linus there would be on Linux. Without Linux the FOSS movement would be in a much worse situation then it is today. And no, I don't think the various BSD flavors are or were a viable alternative. I sailed on that boat and quickly got tired of watching people try to throw each other overboard.
Which come to think of it is very much what seems to be happening right now on this particular issue.
As for git, so far as I know nobody was prevented from working on it for the past three years. The fact that so many preferred to bitch about BitKeeper rather then working on a viable alternative speaks volumes in my opinion.
Is this true? Posted May 3, 2005 17:03 UTC (Tue) by sbergman27 (guest, #10767) [Link] I think you are over-reacting a bit. Linus made a mistake. We all make mistakes; Linus fewer than most. It simply proves that he's human like the rest of us. As to people throwing Linus overboard, I certainly have seen no signs of that. Observing that someone made a mistake is far from throwing them overboard.
Basically, an evecutive mistake was made, there were consequences, which simply happen to have outweighed the extant benefits, and now things are getting back on track. The sky is not falling. Linus is not being burned in effigy. And Git, in the future, will no doubt benefit both the kernel AND the general state of open source SCM projects.
Your comment about not having Linux if it weren't for Linus may be true, but it is irrelevant to the discussion.
Is this true? Posted May 3, 2005 17:11 UTC (Tue) by emkey (guest, #144) [Link] As to people throwing Linus overboard, I certainly have seen no signs of that.Then you haven't read all the responses to this thread. I saw at least one that could easily be characterized in this way. asically, an evecutive mistake was made, there were consequences There are always consequences to any action. The question is, are the reactions in regards to this issue justified? To me that answer is clearly no. Other can and will disagree of course. Your comment about not having Linux if it weren't for Linus may be true, but it is irrelevant to the discussion. Erm, if there were no Linux then we wouldn't be having this discussion at all so logically I don't see how your sentance can be true...
Is this true? Posted May 3, 2005 17:28 UTC (Tue) by sbergman27 (guest, #10767) [Link] >> As to people throwing Linus overboard, I certainly have seen no signs of that.
> Then you haven't read all the responses to this thread. I saw at least one that could easily be characterized in this way.
What in the world are you talking about? Which post?
If it was one of mine, I can assure you that I've no desire to see Linus walk the plank.
At any rate, I fear that this subthread has reached the end of its useful life and is about to wear out its welcome on LWN, which understandably frowns on long "Is Not! Is So!" comment threads.
Is this true? Posted May 3, 2005 17:33 UTC (Tue) by emkey (guest, #144) [Link] Actually the comment I was thinking of was by you...
http://lwn.net/Comments/134450/
In any case, I agree we've pretty much beaten this topic to death for now.
Is this true? Posted May 3, 2005 18:52 UTC (Tue) by sbergman27 (guest, #10767) [Link] I kind of thought that may have been the one. Please read it carefully and try not to read any more into it than is there. (And, yes, I probably overstated the case a bit for dramatic effect.) I do believe that a fork would likely have resulted under those (hypothetical) circumstances. Though, perhaps, it would have been an egcs/gcc style fork.
Is this true? Posted May 3, 2005 19:29 UTC (Tue) by sbergman27 (guest, #10767) [Link] Hihmmm,
emkey disagrees with his interpretation of what I have said. His position is as valid as anyones. At any rate, no need for personal attacks. ;-)
Is this true? Posted May 3, 2005 20:32 UTC (Tue) by emkey (guest, #144) [Link] No, I'm just somebody you disagree with. Life will be much easier for you when you learn to tell the difference.
Is this true? Posted May 5, 2005 20:02 UTC (Thu) by XERC (guest, #14626) [Link] It sounds pretty American:
"Either You are with me, or against
Is this true? Posted May 5, 2005 20:06 UTC (Thu) by XERC (guest, #14626) [Link] OK, may be I was too "poetic".I meant that one should really accept different opinions and not slam the others, who don't agree with one.
Is this true? Posted May 4, 2005 4:26 UTC (Wed) by darthmdh (guest, #8032) [Link] Using BitKeeper was not a mistake. Not silencing the critics enough was.BitKeeper was open source before it came into the limelight because of Linus' adoption for use with the Linux kernel - which was entirely a technical decision as you can go read for yourself in mailing list archives.
A lot of open source SCM people were not very happy about it, because BitMover had done in a short time what they could not do in many years of development. So, in defense of their own egos, they stir up the free software zealots into a frenzy, and otherwise intelligent people (check some of the anti-BK flames by one or two fairly-well-known kernel developers) are reduced to RMS zombies.
It sucks that BitKeeper had to become closed, but understandable given the circumstances. If you ran a grocery store and the guy down the street with a competing grocery store kept coming into yours and nicking items off the shelf to stock his own with, what would you do? Whine incessantly that all groceries should be free? Claim the theif has some kind of right to take your groceries, and how dare you try to stop him?
I think the message that this whole debacle clearly states is that its been well known for more than 3 years that a better SCM tool is required, yet no-one has been able to produce one. And Linus is showing true leadership by, having been forced into a position by the actions of outsiders where he can no longer legally use the perfectly good SCM tool he had, coming up with the beginnings of what may be a worthy piece of free software; instead of bitching and moaning about it like the BK naysayers did for 3+ years.
He's producing good code that is seeing adoption by many other projects also, and its taken, what, a few weeks tops? Have a good think about that - 3 years of bitching and moaning and no code by BK naysayers versus a couple of weeks to get something that is proving good enough to base an SCM on (git is not an SCM, just a filesystem-like layer an SCM can use) by someone who is actually in the open source community and is not simply some politically-inclined freeloading whiner.
Oh, and the comment about throwing overboard appears to be in respect to BSD developers, not Linus. I assume it is in reference to NetBSD forking, hassles over CVS commit access with NetBSD and FreeBSD, etc. In general, the "death of BSD" being due to their internal politics permitting something else where the focus is on development not politics to take marketshare. It's a lesson everyone in the Linux camp must learn from, not be doomed to repeat.
Is this true? Posted May 4, 2005 8:02 UTC (Wed) by xav (subscriber, #18536) [Link] Using BitKeeper was not a mistake. Not silencing the critics enough was.In world are you living ? Free software development isn't your regular corporate environment where a management position allows you to dictate other people's behavior. Devs won't be silenced just because Linus don't want to hear about its mistake. Yes, that was a mistake, plain and simple, and no, just ignoring it isn't the right way to handle it. As always in the community, there will be critics and they will help going forward and avoiding further mistakes. That's the way it always worked.
About technical decisions and groceries Posted May 4, 2005 9:33 UTC (Wed) by man_ls (subscriber, #15091) [Link] which was entirely a technical decisionI wish life was so simple. When decisions involve people, they tend to become political; in this case, the decision to use BitKeeper did certainly involve other people, and was definitely a (negative) political statement -- to the effect of "I don't care about freedom, I just want my process to go smooth". It sucks that BitKeeper had to become closed, but understandable given the circumstances. If you ran a grocery store and the guy down the street with a competing grocery store kept coming into yours and nicking items off the shelf to stock his own with, what would you do?The analogy is flawed; it belongs to the obnoxious category of "mistaking ideas for real objects". Let me rephrase your question: if you ran a grocery store, and the guy down the street keeps stocking the same items that you do (i.e. copying your fine selection of vegetables), what would you do? Don't allow him, any of his family or anyone that has ever spoken with him near your shop? Cover the shop-window and only hand vegetables over the counter? Or just live with it as every merchant on Earth does? instead of bitching and moaning about it like the BK naysayers did for 3+ years.This phrase contradicts your previous statement: yes, BK naysayers did nothing for 3+ years. In fact, they did nothing to copy or clone BK itself; only recently an unrelated developer has seen fit to try and trivially interoperate with BK servers, since the Boss has decided that all your metadata is belong to us. So apparently nobody was even trying to steal the choice of groceries.
Is this true? Posted May 3, 2005 20:26 UTC (Tue) by rickmoen (subscriber, #6943) [Link] emkey wrote:As for git, so far as I know nobody was prevented from working on it for the past three years. In a way, they were: One of the gradually more stringent restrictions BitMover introduced, over time, to the licensing for BK's gratis-usage edition was a non-compete clause, whereby if you or your employer developed, produced, or [re]sold a "substantially similar" competing product, you could not use it. Soon thereafter, in 2004, BitMover also started refusing to sell the full-price edition to developers considered likely to assist in development of competing SCMs. Whether selecting BK was a "mistake" obviously has no objective answer: Plainly, participating kernel developers would say very categorically "No" -- and did so, repeatedly. But, as Neil McAllister points out, it's wise to adopt a proprietary key toolset, if at all, with your eyes open about long-term risks inherent in foregoing control of one's infrastructure, e.g., the pattern of increasing restrictions seen in BK gratis edition's case that hadn't been present when it was chosen (withdrawal of source code access, removal of the convert-to-GPL-if-OpenLogging cases provision, introduction of forced upgrades, introduction of requirement that hosted repositories' source code contents be available on BitMover request via BitKeeper access protocol, no-compete clause, etc.). Rick Moen
Odd that Linus didn't bite at the anticompetition clauses Posted May 4, 2005 0:28 UTC (Wed) by leonbrooks (guest, #1494) [Link] That sort of thing normally gets his goat in an unqualified manner. It'd certainly get mine.
Odd that Linus didn't bite at the anticompetition clauses Posted May 4, 2005 0:44 UTC (Wed) by rickmoen (subscriber, #6943) [Link] Leon:Personally, I found it very surprising when nobody (seemingly) even raised an eyebrow at Larry claiming (during the recent OSDL standoff) that BitMover regarded Linux kernel metadata as BitMover-proprietary. Er, that's the project's change history, folks. That particular bit of vendor overreaching may be academic now, but the example should give people pause, in any similar future situations -- I hope. (BTW, I meant to write "convert-to-GPL-if-OpenLogging-ceases provision", not "cases".) Rick Moen
Odd that Linus didn't bite at the anticompetition clauses Posted May 4, 2005 1:12 UTC (Wed) by daniel (subscriber, #3181) [Link] "Personally, I found it very surprising when nobody (seemingly) even raised an eyebrow at Larry claiming (during the recent OSDL standoff) that BitMover regarded Linux kernel metadata as BitMover-proprietary. Er, that's the project's change history, folks."
Nobody? Did you ask me?
Regards,
Daniel
Git couldn't have happened before now Posted May 4, 2005 13:45 UTC (Wed) by kevinbsmith (guest, #4778) [Link] "As for git, so far as I know nobody was prevented from working on it for the past three years."
Honestly, if *anyone* other than Linus had invented git, three years ago or last month, it would not have been noticed. So the fact that Linus was pretty committed to BK did, in fact, have a suppressive effect on SCM development. It didn't shut it down completely of course--arch, monotone, darcs, and others went from unusable to fairly mature during those three years.
20/20 hindsight Posted May 3, 2005 17:00 UTC (Tue) by karath (subscriber, #19025) [Link] I see a number of problems with this analysis.
First: Linus' views on source control have undergone radical change in the past 3-4 years. Until 1 or maybe 2 years ago, he did not have the knowledge base or processes around him to understand enough to build git.
Second: Source control is not his first, second or even third love in life... Only under severe pressure - like BK being pulled from under him - would he have been motivated to go and do something.
Third: Linus was fairly free with information on his requirements of an SCM over the past 3 years. Why did none of the free SCM developers take him at his word and go do something that would have met his requirements. Perhaps (I am speculating here) he hoped that by the time a further BK decision was forced upon him, one of them would be ready. And he did some inspection before starting git - Monotone certainly got a major boost from his attention.
regards,
20/20 hindsight Posted May 3, 2005 17:09 UTC (Tue) by sbergman27 (guest, #10767) [Link] Well. I was one of the (many) people predicting an end result very much like this 3 years ago. My foresight is not the best, but I'm not blind either.
BTW, I will agree with your first argument. I almost included a mention of the possibility in my first post, but decided to let someone else bring it up, if they wanted to. Much of the benefit of using Bitkeeper was not due to the tool itself, but what the kernel developers learned from using it. Fortunately, these are lessons they can take with them. No one denies that BK is a high quality tool. However, many feel that it's use was inapproprate and doomed to failure.
Is this true? Posted May 3, 2005 16:57 UTC (Tue) by pflugstad (subscriber, #224) [Link] BitKeeper allowed the Linux kernel to make huge strides the past few years.Actually, I think that the process that BitKeeper required Linus and company to follow is where the huge strides were made. The fundamental problem when this whole thing started was the Linus didn't scale: he coulnd't keep up with the flow of patches coming in. BitKeeper essentially forced him to start trusting other developers and that's why we're seeing 10MB/month merge rate for patches today. I expect that any SCM tool, distributed or not, would have basically forced the same result. Say Linus had used CVS - there would be a central repository and there would have been Linus with commit capability. Still the same problem, Linus doesn't scale. Until he opened up and let others essentially commit changes (which is what BitKeeper forced, albeit in a distributed manner). That's where the real progress has came from.
Is this true? Posted May 4, 2005 14:01 UTC (Wed) by kevinbsmith (guest, #4778) [Link] "I'm sorry, I still don't see what Linus did as a mistake."
Knowing what we now know, I believe that using BK was a net positive for kernel development. It helped the process, and taught Linus a lot about version control. Remember that prior to BK he didn't use any tool at all. Larry and the BitMover folks have also been EXTREMELY gracious in how they have handled this termination, and Linus has done a great job getting git working so quickly.
HOWEVER, three years ago, when the choice to use BK was made, there was no way to predict that things would turn out so well.
There was no way of knowing how long it would take for Larry to pull the plug, nor how productive Linus would actually be using BK. There was no way of knowing whether Larry would suddenly shut down the server and yank all the licenses with no warning, or otherwise try to prevent access to the project history. There was no way of knowing that Larry wouldn't sue Linus for developing something like git. There was no way to know how long it would take for git to become a productive tool (few people would have guessed "less than a month"). In fact, several of these outcomes are quite surprising, and would not have been predicted by most people.
So if just a couple things had turned out differently, this could have been a big net loss. Thus, *at that time*, choosing BK was very risky, and therefore was, in my opinion, a mistake.
I'm just glad everything did work out well, and I really appreciate how Larry has handled everything after he decided to pull the gratis license.
|
|
||||||||||||