Well, you forgot a step. You have to shop the vaporware around to get funding to build said "product". And then, if you do a good job, you get to build a product. Either that, or you have lots of money, lots of spare time, or both.
I beg to differ. This is why I haven't started my own buisness yet (this and being risk-adverse). I would spend too much time messing around with business crapola, which is what I don't like. I just don't see how you could have a successfull small business without selling your "product".
The only reason I can (mostly) do this ("deep code" and stay in my cubicle all day) is there is someone down the hall selling our "product".
yeah, except compiled code is by definition compiled on the target platform, so therefore, more modern compilers can optimize based on the platform you will run it on.
Can't help but wonder if the present generation of "faster, better, cheaper" probes will ever live this long though.
Interesting point, but do they need to? How long is long enough. Is it a waste of money/time to build it to last for 30 years?
For near earth stuff, it would make more sense to build it just good enough and save the money for the next project. For far flung stuff (like voyager and pioneer), is the data useful? If not, what a waste of resources.
While you make some good points, you are being overly harsh. This guy, while maybe a bit misguided, had some good things to say. It is a viewpoint. You disagree. Fine. Maybe he has bought into something that you find suspect, fine too. Give the guy a break.
Finally, your presentation is not credible. Your post contains zero grammar or spelling errors, no technological jargon or acronyms, and multiple marketing buzzwords. You are obviously not a programmer or a sysadmin; your background is clearly marketing and public relations.
Kidding? I hope so. If not, you are kidding yourself.
I am very concerned that this post was moderated to 5. This means one of two things: either PR people are not just posting to Slashdot but also moderating, or the average Slashdot moderator is unable to recognize PR rhetoric when they see it. I submit this message in the hopes of helping to fix the latter if it is the case.
What are you trying to say, that we only support one viewpoint here? and that since something else is posted that it is suspicious? That is total crap.
Perhaps I will get modded down for this, but I think you need a nap or something. While you have a good point, saying that this viewpoint has no place in the discussion here is just crap.
They could today if they wanted to. Jabber has tried this and got it to work, but AOL and MSN, etc. kept changing their protocol to break Jabber's integration. They even went so far as to make the protocol very dynamic, in order to ease their constant changing of protocols. Very conter-productive if you ask me.
If they wanted to, they could, but then, that would hinder AOLs big selling point: all your friends are on it. Then the Internet came, and it didn't matter. They hold onto what they can and lie about the rest.
Didn't mean to imply that there was one disk per 100,000 servers. What I've seen/done is group the servers into "farms" and share disk between the "farm". Each farm having 30-100 servers, depending on the application.
disk is cheap
but administration is expensive. It takes too long to try to keep 100,000 or even 100 server upto date or in sync. There just isn't any easy way.
Once again, drawing from my experience, have only the OS and swap/tmp space on local disk, then each farm shares a NetApp or EMC2 type disk array (or two or three depending on locality). These are machines that have more than 99.99% uptime. Also, you can point a farm to different fileservers within a minute, so there is your redundency. Also, you can mirror (NetApp calls it "snap mirror") the fileservers, making one a master and the rest slaves, so you update one and give a command, presto, all of the others are updated within 10 minutes.
I'm not saying that is the only way to do it, but that is the only the "big boys" are doing it.
Actaully, this probably shouldn't be considered a cluster, per se. I imagine they use load-balanced clones with shared disk. It is the only way I have ever seen to run more that about 50 web servers without nightmares seen only in IBM or MS commertials.
This is very different of the "beowulf" type of cluster. There isn't any shared computation going on, since each request would be very minor by itself. I assume they have that many boxes because of the number of connections at a time.
note that this comment comes from my experience in the industry, not with Google itself.
windows will never truly be able to escape the limitations of the x86, since if you are going to lose all your legacy software you might as well dump windows altogether.
very nice.
But linux, all that legacy software can be freely compiled
Almost. I would like to see a working version of xtank for linux. The source doesn't compile, and I haven't found binaries that worked. Oh, well, other than that, I can't really complain.
yeah, that would be the "IBM sells said HD activities (except R&D)" part.
Re:This is arguably *the* most critical problem
on
Version Fatigue
·
· Score: 2
Perhaps this is true, it sounds pretty good.
However, you could make the argument that the MS libraries are thus better and better, with each revision. You could also say that Xlib sucked in the 70s and it sucks now; that Motif sucked in the 80s and it sucks now. Some say this is why "linux on the desktop" is not "done". If the UNIX environment could get a new graphics library or even a new way of thinking, much could be done to improve things. If we didn't have to constantly have to work around stuff that we learned was bad 20 years ago, we could make things better.
I don't totally agree with this, but it has been said many times before, and I can't totally disagree.
My view is that there is a happy medium between having stuff that has been around for 40 years and stuff that is constantly changing (read: improving). Yes, it is good that my cumlitave experience on UNIX is mostly applicable, but it would be nice if more improvements were possible. But I don't think it is good to change too much either; you get a bunch of people that can't possibly have more that 2 years experience in anything that matters.
And yes, I did accedentally forgot to turn off the +1 thing, but really, what difference does it make. It's not like I am really moding my own post or anything.
Nonsense is exactly what the post was, perhaps you took it too seriously. It was never intended to convey any real info or a real opinion. I just thought it was funny, especially given that whenever I want to find a piece of software, I usally have to make do with an alpha or beta version that I have trouble compiling (yes, I am a programmer, but there are limits to the amout of time I want to spend). As if the US government would just find "ControlTower 1.0" out of sf or fm. For all of OSS's benifits, finding finished, usable software consistently isn't one of them.
I wonder about the "mass removal of moderation priveleges". There seems to be too much noise for it to be a misunderstanding. I went a year or so without a moderation prompt, but then it appeared. Strange. I figure they do the best they can to keep the signal to noise ratio up as high as possible, but they seem to sometimes go too far. Whatever.
While I pretty much agree with this, my answer is: "well sorta".
There is a considerable fraction of the business managers in the world that would say that the Weather Channel just paid for driver development for all their competitors. That is, their competitors will now have the same drivers (some of which need them just as much as TWC), but they didn't have to spend the time or the money on it. Thus, they now have a competitve advantage.
I think that this is somewhat backwards thinking, but I can't totally disagree either. Also, most people's bosses wouldn't disagree, and that is a major issue.
Slashdot Mirror
Idiot. Password123! is the way to go. I thought everyone knew that.
microsoft: 2004 called and we want our IDE back.
translation: buy a new one?
Using a modern version of Word, output in WordML (xml format). Use a XSL stylesheet to convert the WordML to FO (formatting objects).
From there, do anything you want, like XHTML or PDF.
Or just go to XHTML from WordML with some stylesheet. XSL is teh cool!
Well, you forgot a step. You have to shop the vaporware around to get funding to build said "product". And then, if you do a good job, you get to build a product. Either that, or you have lots of money, lots of spare time, or both.
I beg to differ. This is why I haven't started my own buisness yet (this and being risk-adverse). I would spend too much time messing around with business crapola, which is what I don't like. I just don't see how you could have a successfull small business without selling your "product".
The only reason I can (mostly) do this ("deep code" and stay in my cubicle all day) is there is someone down the hall selling our "product".
I wish/hope I am wrong.
When I was in school, I remember thinking that common lisp wasn't very common at all. (me so funny)
Now, this article made me think...
Practical Common Lisp. Lisp is neither Practical, nor Common.
Oh yeah, I still crack myself up....
Can't wait until power over wireless connections become available.
then you'll really have something.
yeah, except compiled code is by definition compiled on the target platform, so therefore, more modern compilers can optimize based on the platform you will run it on.
Sorry, try again.
I know how to do this, I have been doing this for so long, I can't believe you guys are still searching for the right way.
I have it.
I box it up and put it in my guest room closet. Problem solved.
The one on the right looks a lot like a picture I took of my dog. Scary.
I just don't remember that background....
For near earth stuff, it would make more sense to build it just good enough and save the money for the next project. For far flung stuff (like voyager and pioneer), is the data useful? If not, what a waste of resources.
Can you say "solution in search of a problem"?
One of those "our school is better than yours" type of things, I guess.
While you make some good points, you are being overly harsh. This guy, while maybe a bit misguided, had some good things to say. It is a viewpoint. You disagree. Fine. Maybe he has bought into something that you find suspect, fine too. Give the guy a break. Kidding? I hope so. If not, you are kidding yourself. What are you trying to say, that we only support one viewpoint here? and that since something else is posted that it is suspicious? That is total crap.
Perhaps I will get modded down for this, but I think you need a nap or something. While you have a good point, saying that this viewpoint has no place in the discussion here is just crap.
The answer is: not until they want to.
They could today if they wanted to. Jabber has tried this and got it to work, but AOL and MSN, etc. kept changing their protocol to break Jabber's integration. They even went so far as to make the protocol very dynamic, in order to ease their constant changing of protocols. Very conter-productive if you ask me.
If they wanted to, they could, but then, that would hinder AOLs big selling point: all your friends are on it. Then the Internet came, and it didn't matter. They hold onto what they can and lie about the rest.
Didn't mean to imply that there was one disk per 100,000 servers. What I've seen/done is group the servers into "farms" and share disk between the "farm". Each farm having 30-100 servers, depending on the application.
disk is cheap
but administration is expensive. It takes too long to try to keep 100,000 or even 100 server upto date or in sync. There just isn't any easy way.
Once again, drawing from my experience, have only the OS and swap/tmp space on local disk, then each farm shares a NetApp or EMC2 type disk array (or two or three depending on locality). These are machines that have more than 99.99% uptime. Also, you can point a farm to different fileservers within a minute, so there is your redundency. Also, you can mirror (NetApp calls it "snap mirror") the fileservers, making one a master and the rest slaves, so you update one and give a command, presto, all of the others are updated within 10 minutes.
I'm not saying that is the only way to do it, but that is the only the "big boys" are doing it.
Actaully, this probably shouldn't be considered a cluster, per se. I imagine they use load-balanced clones with shared disk. It is the only way I have ever seen to run more that about 50 web servers without nightmares seen only in IBM or MS commertials.
This is very different of the "beowulf" type of cluster. There isn't any shared computation going on, since each request would be very minor by itself. I assume they have that many boxes because of the number of connections at a time.
note that this comment comes from my experience in the industry, not with Google itself.
omg. Its been so long, I have forgotten that little bit. Brings back some memories. Very nice.
Well, there's the first part anyway, you get the idea.
yeah, that would be the "IBM sells said HD activities (except R&D)" part.
Perhaps this is true, it sounds pretty good.
However, you could make the argument that the MS libraries are thus better and better, with each revision. You could also say that Xlib sucked in the 70s and it sucks now; that Motif sucked in the 80s and it sucks now. Some say this is why "linux on the desktop" is not "done". If the UNIX environment could get a new graphics library or even a new way of thinking, much could be done to improve things. If we didn't have to constantly have to work around stuff that we learned was bad 20 years ago, we could make things better.
I don't totally agree with this, but it has been said many times before, and I can't totally disagree.
My view is that there is a happy medium between having stuff that has been around for 40 years and stuff that is constantly changing (read: improving). Yes, it is good that my cumlitave experience on UNIX is mostly applicable, but it would be nice if more improvements were possible. But I don't think it is good to change too much either; you get a bunch of people that can't possibly have more that 2 years experience in anything that matters.
Yeah, real hilarious.
And yes, I did accedentally forgot to turn off the +1 thing, but really, what difference does it make. It's not like I am really moding my own post or anything.
Nonsense is exactly what the post was, perhaps you took it too seriously. It was never intended to convey any real info or a real opinion. I just thought it was funny, especially given that whenever I want to find a piece of software, I usally have to make do with an alpha or beta version that I have trouble compiling (yes, I am a programmer, but there are limits to the amout of time I want to spend). As if the US government would just find "ControlTower 1.0" out of sf or fm. For all of OSS's benifits, finding finished, usable software consistently isn't one of them.
I wonder about the "mass removal of moderation priveleges". There seems to be too much noise for it to be a misunderstanding. I went a year or so without a moderation prompt, but then it appeared. Strange. I figure they do the best they can to keep the signal to noise ratio up as high as possible, but they seem to sometimes go too far. Whatever.
More like version 0.1.4 BETA.
I rarely see a version 1.0 from sourceforge or freshmeat or anything like that.
While I pretty much agree with this, my answer is: "well sorta".
There is a considerable fraction of the business managers in the world that would say that the Weather Channel just paid for driver development for all their competitors. That is, their competitors will now have the same drivers (some of which need them just as much as TWC), but they didn't have to spend the time or the money on it. Thus, they now have a competitve advantage.
I think that this is somewhat backwards thinking, but I can't totally disagree either. Also, most people's bosses wouldn't disagree, and that is a major issue.