eye-opener
|
| Author | Content |
|---|---|
| grouch Jul 27, 2006 11:47 PM EDT |
Corbet says, "The talk is very much worth a read." That's understating it by a mile. I'm on an enfuriatingly slow dial-up connection and it was worth waiting for each and every slide in Greg Kroah-Hartman's presentation. It's just fantastic. |
| Sander_Marechal Jul 28, 2006 1:35 AM EDT |
Just wait patiently for that very last slide :-) Also, make sure you read stable_api_nonsense.txt referred to in the slides (link: http://developer.osdl.org/dev/robustmutexes/src/fusyn.hg/Doc...). A great read as well. |
| grouch Jul 28, 2006 8:57 AM EDT |
Part of the reason for that last slide is revealed 'way back at the start: "Yes, that's right, we support more things than anyone else. And more than anyone else ever has in the past." |
| dek Jul 28, 2006 10:29 AM EDT |
To the editors: I posted that there Kroah - Hartman talk to LXer when I read it!!! I kept looking for it here in the news page thinking that someone would recognize its brillance but --- NOOOOOO!! Sheesh!!! (Just giving you guys a hard time!!!) ;-) Don K. |
| grouch Jul 28, 2006 10:48 AM EDT |
dek: I searched through the 300+ stories that were rejected between now and Jul 20 (before the keynote) and didn't find any with "Kroah" in them. (That's about as unique a word to this story as I could think of). I don't know what happened. |
| dek Jul 28, 2006 12:39 PM EDT |
grouch: Check my lxer page. It's listed there. Maybe I'm not understanding something correctly?? If so, I'd like to know . . . . Don K. |
| grouch Jul 28, 2006 1:14 PM EDT |
dek: I'm the one who is not understanding. I went looking through the deleted stories because you said, "I kept looking for it here in the news page thinking that someone would recognize its brillance but --- NOOOOOO!!" However, when I look at your page I see the title, "Myths, Lies, and Truths about the Linux kernel", which appeared on the Newswire on July 26, posted by you. I don't understand the complaint -- how can it not show up on the "news page" and yet also have been posted? |
| dek Jul 28, 2006 1:56 PM EDT |
My apologies, grouch. I see that it was approved and did appear in the newswire. Now, I'm not sure that I understand the distinction between the newswire and LXer Linux News. I had thought that this would go on the LXer Linux News front page which is where I was looking for it. I completely missed the newswire. Pardon my confusion. Maybe a word of clarification within the story submission guidelines as to the distinction between the two (if there is any) would be helpful here. Don K. |
| grouch Jul 28, 2006 5:56 PM EDT |
dek: The Newswire is just that big column of news scrolling down the front page. Your submission showed up on the front page in the middle of the night, U.S. time. With approximately 2 stories popping out at the top of the page every hour, it would have scrolled off the bottom in about half a day. Your submitted story is part of the database, accessible at http://lxer.com/module/newswire/view/65830/index.html BTW, "Ottawa Linux Symposium 2006 Keynote Presentation about the myths and truths of linux kernel development" is even more understated than Corbet's description. Neither gives a glimpse of the sledge-hammer blows Kroah-Hartman used against those myths. ;o) |
| dek Jul 29, 2006 8:44 AM EDT |
I feel it is best to understate the case!! That way people have the opportunity to feel pleasantly surprised. ;-) Grouch, I thank you and hope you understand my confusion. For future reference, is it necssary to have two newswires repeating each other on the same page? Seems a little redundant . . . . Sorry to have hijacked this thread. I hope we can continue on with the discussion of Kroah-Hartman's presentation. Don K. |
| sbergman27 Jul 29, 2006 9:04 AM EDT |
The biggest eye-opener from the symposium that I have yet encountered is the last paper from volume 1 of the proceedings: http://www.linuxsymposium.org/2006/linuxsymposium_procv1.pdf It starts on page 441 of the pdf. Read it and then consider how it can be reconciled with some of the "truths" we hold so dear: 1. With many eyeballs, all bugs are shallow. 2. Open source code quality is better than proprietary due to better peer review. 3. Open source security is better because anyone can see the flaws and report them. Dave presents many examples of FOSS code doing egregiously improper things that even the authors didn't know about. And it's not like these are obscure programs that get little use or attention. And if these, relatively simple code bases have these problems, I shudder to think what could lie undiscovered in Firefox and OO.o. I've heard this presentation refered to as "humorous". I don't find it funny at all. |
| dinotrac Jul 29, 2006 9:27 AM EDT |
Steve - Better is better, not perfect. Still, some of this, I believe can be traced back to Computer Science programs that preach some nonsense about how it is bad to do "premature optimization." Of course, that inevitably gets carried to, "well, everything is so fast these days, a cycle here or a cycle there won't matter" which leads to simply not paying any attention. You know where that leads. |
| jdixon Jul 29, 2006 9:40 AM EDT |
> And if these, relatively simple code bases have these problems, I shudder to think what could lie undiscovered in Firefox and OO.o. Steve: What Dino said. :) Look at it another way. If the open source code (freely available to all to review), has so many problems, think how bad the closed code (with a much more limited review) must be. |
| sbergman27 Jul 29, 2006 10:07 AM EDT |
> Look at it another way. If the open source code (freely available to all to review), has so many problems, think how bad the closed code (with a much more limited review) must be. I'm not ready to just discount the embarrassing evidence like that. After all, we're not talking about a few cycles here or there. We're looking at long-standing pathological behavior in ubiquitous FOSS apps which apparently no one has noticed in... how long? 'Perfection' isn't exactly in this neighborhood. At any rate, this is something that can be tested. There is really not much that Dave did that required the source code. Strace, ltrace, and systemtap work just as well on closed source binaries as open. Closed source does not completely obscure black box testing. For example, when Michael Zalewski got the radical idea of actually *testing* how FOSS's most popular desktop app and security darling actually handled unexpected (random) input ( http://tinyurl.com/43dm8 ) he found a similarly embarrassing situation in which FF blew up in seconds or minutes whereas IE could handle hours of testing without suffering potentially exploitable crashes. Took the gloat out of my sails for a while. Then again, maybe they don't teach basic input validation any more, either. |
| dinotrac Jul 29, 2006 10:16 AM EDT |
Steve - I certainly don't mean to diminish the importance of good design and validation for free software projects. I very certainly do not defend laziness in that regard. Wish I knew how many times I've posted (sometimes with very nasty flames as my reward) that we should not get complacent, that there is nothing about free software or linux or anything that makes us immune to bugs, stupidity, or vulnerability. Let us not forget... Some of the best software ever written was written by employees of companies writing closed-source proprietary software. Remember Y2K? At least part of that problem came about because there were mission critical systems out in the world that had been running with relatively few changes for thirty or more years. No -- of course I don't mean continuously. But in this era of constant re-factoring (I sometimes think initial design should just be called pre-factoring or preliminary-re-factoring), that is pretty mind-boggling. There is no substitute for good practices applied by conscientious people, whether the source is open or closed. I think open source has additional benefits, but it still depends on quality people. |
| grouch Jul 30, 2006 1:45 AM EDT |
dek: >"For future reference, is it necssary to have two newswires repeating each other on the same page?" Ok, I'm lost again. What two newswires? There is the Newswire (main, wide column) and the RSS version with just title, date and discussion count (small, right-hand column). |
| dek Jul 30, 2006 5:54 AM EDT |
Hi grouch,
I'm referring to having the RSS feed on the same page as the Newswire itself. Generally just the RSS link is enough and any more seems redundant. Don K. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!