nope, it was a great place for an open source project to grow, and a lot of folks there supported it. it was one of the first online magazines to go with a full opensource back end.
Well I guess my point wasn't obvious. Sorry about that...let me see if I can make it clearer.
My point was in response to this:
"Sorry, but blog evangelists have no credibility among those who like to use their brain when viewing news."
Which I took to mean that basically bloggers have no credibility. My point is that bloggers do and can have credibility, in the way that Om Malik points out.
A blogger who blogs about something that he actually works on/with, doesn't have as much credibility as someone who is an independent observer of the product, and has knowledge or history with the product. That's my Linus Torvalds/lizard researcher analogy.
Jeremy zawodny works at yahoo, and blogs about products at yahoo, and yet still remains to be credible in the face of that because he isn't afraid to be critical of yahoo's products, and even praising of other companies' products (i.e. Google Maps, etc.)
So I'm agreeing with Om Malik about blogger credibility, and disagreeing with anyone who blanketly declares that bloggers have no credibility.
if you've been around silicon valley, money is not all that hard to get, compared to getting/developing a product that actually *deserves* to be funded.
Heiki Tuuri has been talking quite a bit about instability of InnoDB on AMD opterons, especially on 2.6.
The fact is - it's just plain unpredictable and IMHO, broken. Andrew Morton has been given ample evidence of this, and it certainly looks like the majority of the problem lies in either virtual memory, or the 2.6 I/O scheduler.
Friendster suffered the same problems, and came to one conclusion: use 2.4, until 2.6 is fixed.
My current work shows the same thing: even a CHECK_TABLE would fail with "page corruption", and it would appear as if it's random.
FWIW, there's even a report on the mysql mailing list of a guy running his own test...he memory-mapped an entire 4gb or so, and got back different results when it was re-read. Very scary.
Indeed, I think many who are posting on this aren't quite aware of what these packages are intended to provide.
Having much more experience with the FEM and madymo packages myself, I haven't had any chance to testify in any accident *reconstruction* capacity (my background was in research).
I assume that most of the testimony is usually about backing up claims of damage based on trauma ? and if so, at what level of detail is the testimony ?
is it just whether or not a certain amount of trauma *could* have occured, in certain configurations ?
PC-Crash, FWIW, is not a *real* vehicle crash simulation package, IMHO. validated and comphrehensive results can be obtained with simulations, and do, every day.
the results of those validated simulations go into the NHTSA rulemaking body, which gets all of the compliance rules together so auto manufacturers aren't allowed to sell cars with tinfoil frames.
yeah, the work done by LSTC's Dyna-3d, Altair's HyperMesh, and MSC'S Patran can give basically everything that you could *ever* want to know about explicit dynamic phenomena.
I did about 4 years working for the Vehicle Crashworthiness Division for the US DOT using the above programs, and they are quite complex and used for mostly research, not accident reconstruction.
generally speaking, all of this (temperature, materials, pressure, etc) can, and does, get calculated with *real* finite-element analysis of vehicle crash simulations.
PC-Crash is not one of those pieces of software, however.
Slashdot Mirror
nope, it was a great place for an open source project to grow, and a lot of folks there supported it. it was one of the first online magazines to go with a full opensource back end.
Well I guess my point wasn't obvious. Sorry about that...let me see if I can make it clearer.
My point was in response to this:
"Sorry, but blog evangelists have no credibility among those who like to use their brain when viewing news."
Which I took to mean that basically bloggers have no credibility. My point is that bloggers do and can have credibility, in the way that Om Malik points out.
A blogger who blogs about something that he actually works on/with, doesn't have as much credibility as someone who is an independent observer of the product, and has knowledge or history with the product. That's my Linus Torvalds/lizard researcher analogy.
Jeremy zawodny works at yahoo, and blogs about products at yahoo, and yet still remains to be credible in the face of that because he isn't afraid to be critical of yahoo's products, and even praising of other companies' products (i.e. Google Maps, etc.)
So I'm agreeing with Om Malik about blogger credibility, and disagreeing with anyone who blanketly declares that bloggers have no credibility.
what does google do that is truly unique on the web ?
I can't think of anything that they do that is truly unique, and all of which, yahoo does.
if Google is run by engineers, then why don't they have:
-creative commons search
-rss on a my.google.com page
-rss everywhere
-an image search that actually works
-traffic overlay onto their maps
Would you ever buy a book about linux by Linus Torvalds ?
Would you ever watch a TV broadcast about lizards hosted by a guy who has studied lizards all his life ?
Then why don't you think some bloggers can have credibility ?
fortnight. sure thing.
try getting something printed in a newspaper, without being a journalist who works there.
blogs pointed out the Kryptonite lock fault, and it cost the company $10 million in lost revenue.
In 10 days.
some fad, that is.
I think this is such a great thing!
orkut is very much overrun with Brazilians, more than anyone else.
if you've been around silicon valley, money is not all that hard to get, compared to getting/developing a product that actually *deserves* to be funded.
whoops...replied to the wrong post.
how will Yahoo be selling the work of developers who use their API ?
I'm not sure I see your point.
how will Yahoo be selling the work of developers who use their API ?
I'm not sure I see your point.
Heiki Tuuri has been talking quite a bit about instability of InnoDB on AMD opterons, especially on 2.6.
:)
The fact is - it's just plain unpredictable and IMHO, broken. Andrew Morton has been given ample evidence of this, and it certainly looks like the majority of the problem lies in either virtual memory, or the 2.6 I/O scheduler.
Friendster suffered the same problems, and came to one conclusion: use 2.4, until 2.6 is fixed.
My current work shows the same thing: even a CHECK_TABLE would fail with "page corruption", and it would appear as if it's random.
FWIW, there's even a report on the mysql mailing list of a guy running his own test...he memory-mapped an entire 4gb or so, and got back different results when it was re-read. Very scary.
Either way, I sympathize with you.
just out of curiosity...it wasn't 2.6.x kernel running on Opterons, was it ?
and perhaps mysql using O_DIRECT ?
Indeed, I think many who are posting on this aren't quite aware of what these packages are intended to provide.
Having much more experience with the FEM and madymo packages myself, I haven't had any chance to testify in any accident *reconstruction* capacity (my background was in research).
I assume that most of the testimony is usually about backing up claims of damage based on trauma ? and if so, at what level of detail is the testimony ?
is it just whether or not a certain amount of trauma *could* have occured, in certain configurations ?
those can be taken into account by FEM analysis, too. :)
just, PC-Crash doesn't.
indeed that is true.
PC-Crash, FWIW, is not a *real* vehicle crash simulation package, IMHO. validated and comphrehensive results can be obtained with simulations, and do, every day.
the results of those validated simulations go into the NHTSA rulemaking body, which gets all of the compliance rules together so auto manufacturers aren't allowed to sell cars with tinfoil frames.
yeah, the work done by LSTC's Dyna-3d, Altair's HyperMesh, and MSC'S Patran can give basically everything that you could *ever* want to know about explicit dynamic phenomena.
I did about 4 years working for the Vehicle Crashworthiness Division for the US DOT using the above programs, and they are quite complex and used for mostly research, not accident reconstruction.
A trillion simulations are a helluva lot cheaper than 1 full-scale test.
generally speaking, all of this (temperature, materials, pressure, etc) can, and does, get calculated with *real* finite-element analysis of vehicle crash simulations.
PC-Crash is not one of those pieces of software, however.
sucking or not, they have a better stock price than VA Linux does (i.e. slashdot).
like Yahoo, when they use the "POS" database ?
we've been trying it since March:
here's the issue: (O_DIRECT)
http://lkml.org/lkml/2004/10/22/19
if you think Yahoo only does 5k requests per second, then you'd be mistaken.
Did you try load balancing Innodb slaves ?
btw, mysql on opterons is quite excellent, but don't even think of using 2.6.xx kernels on AMD64. just fyi, it was pretty awful when we tried it.