People develop for Linux just because that's what they (we) are using.
I've come across the opposite problem: I wrote a daemon on Linux, announced it asking for feedback especially on non-Linux systems (BSD, Solaris, etc), and got exactly none. And it was not for lack of interest; I did get a fair number of downloads. So my deamon is still in the state of "works fine under Linux, you're on your own elsewhere".
Oh well, one of these days I'll have to install FreeBSD on a spare partition...
Never was the saying, "a picture is worth a thousand words" more true
than on the Internet.
yeah, especially in the number of bytes it takes. now, when it comes to value and interest, it's more like a word is worth half a dozen.gifs. just do a "page info" from netscape on any page on a commercial website, and go through the list of.gifs and.jpgs... most of them are completely uninformative, they're just put there to make the layout "cute" and the page slower.
I had the *exact same feeling*. Ender's Game was a good, fast and compelling read. Speaker for the Dead was an amazing broadening and deepening of both the subject and the characters. Xenocide had great quotes, great characters (though both taoism and puritanism get caricatured rather badly), great situations, and a shitty deus-ex-machina ending that fit like a soccer ball in the soup. for that reason I haven't read Children of the Mind (or whatever the following sequel is called), although I'll probably pick it up if I see it cheap enough in a 2nd hand bookstore. As for the 'shadow' series, I really don't find the premise all that compelling, nor the characters of Bean or Peter. Give me Valerie's story anytime, and I'll buy it... no wait, no need, it's already in Speaker for the Dead.
Making a good dynamic site is more complicated than sticking some php/java/vbs/whatever code in the middle of an html file. First, you really don't want to put your programming logic and your html in the same file, for two reasons: 1) programmers and designers should not interfere w/ each other's work, and 2) program code stuck in the middle of text is not maintainable; for large applications you really want to create libraries of functions and do things in a structured way. Second, you probably want to treat your whole website as an OO application, to handle a hiearchy of common pieces (ex. navigation) as well as looks or themes.
There are *many* technologies out there to do these things right. Apache with mod_perl and either AxKit or the Templating Toolkit (or others) can do it. I hear PHP can do it using the right libraries. Java servlets, for all their verbosity, can do it using something like go.com's Tea system.
I have never used ColdFusion or ASP myself, but I know many people who
have, and I've repeatedly heard that they're great for quickly building something, but very bad for managing complexity.
Is it me, or did Micros~1 just admit in public that Linux has considerable advantages? I mean, how else could they be saying that it doesn't *only* have advantages?
hell, redhat PERIOD, *shudder*. it comes with more holes than a
sieve by default...
how long has it been since you installed a copy of RedHat? there are many kinds of installs these days, and it doesn't start all kinds of network services in all installations anymore. compared to, say, RH5.x, they've been getter quite a bit more secure. then again, if you're installing a server, you're supposed to know what a listenign service is, how to turn it on or off, and whether you should have it on or off. it's not like UNC's sysadmins are clueless cable-modem newbies.
bash MS all you
want, but W2k (and NT) have MANY less opportunities to be owned
remotely, you have to give them that.
then how is it that more than half the website defacements are on IIS servers, while only about 30% of webservers run it?
Re:Should be possible but by an algorithm?
on
Does P = NP?
·
· Score: 2
I take it that you're saying you don't believe in Turing's thesis. could you specify a bit more what you mean by "intelligent way" as opposed to "analytical way"? specifically, do you see this 'intelligent way' as being specifiable by a good old formal-language algorithm?
The popular spots in the Himalayas *are* getting worryingly filthy. I haven't been in tha Everest area (which btw isn't that hard to access; trekking to Everest Base Camp is hugely popular, it's only going up further that gets hard), but I've been in the Annapurna range, and I was apalled at the amount of junk that litters the paths there. The camp at Thorung Phedi (last stop before the Thorung La, at 4100m or so) is quite dirty. Of course, the view from there more than compensates for it;)
oh yeah? and going from horses to cars has made a "new world" just how? it's the same old world with the same old deep problems, except that we can move things (and people) around considerably faster now. same with computers, we now have them all over the place and sure they've changed a lot of how we spend our hours (most of us/.ers should know, being paid to spend our days in front of computer screens). yet I see no evidence that the introduction of computers deserves anything like the name of a "new world". it's easy to cry "revolution" when you're in the middle of it.
i have nothing against technology, in fact I love to play with it. but I'm deeply convinced that the major bottlenecks in human society and evolution are *not* technical, which means that technology isn't going to solve them, nor (I hope) to make them substantially worse. which is fine, really, as it's not its job. computers are all about easily doing things that we could already do, but were (hugely) prohibitive in cost and time. what gets branded as "innovation" from the computer world (and I don't only mean MS here) is just ridiculously simple compared to the real human problems that each of us has to grapple with, and I see no evidence at all of that changing.
If the guy needs an elaborate disclaimer just to say that he doesn't empathise with computers, and then states that his position is going to be "unpopular and resented by his professional and social environment".... all I can say is that his social environment must be a really warped corner of the world where common sense is singularily scarce.
seriously, most of what he says makes so much sense that it's scary. there have been documented cases of end of the world theorists since at least 2000 years ago, and people still believe it when new ones come along with new and shiny "singularity" scenarios...
that's one side of the story. the other side is that clone() doesn't support the exact POSIX threads API, and the linuxthreads library has to bend over backward (and be slow, in the process) to add this support. the situation is slowly fixing itself, but there are some human communication problems in the middle, and it takes a lot of time for a new version of linuxthreads to make it all the way to the distributions.
that *still* depends on the workload. if it's CPU-bound, it just works. if it's I/O bound, it depends on global system recourses (bus speed and bandwith), and on how well smp-threaded that particular subsystem is on the OS. one of the big improvements of the linux 2.4 kernel is precisely a major overhaul of the tcp/ip stack to smp-thread it completely. OTOH, heavy filesystem work (lots of rename(), unlink(), open(O_CREAT), etc) is not likely to scale super-well, because of filesystem locking constraints, and that's under all OSes too, which is why you generally avoid writing your big programs in such a way that they have to be messing around with lots of little files all the time (e.g that's why you use a database backend, or some kind of db format, rather than flat files, if you want a news server that scales). beyond that, in the real high end (32 procs and more), the SMP model itself starts showing its limits, which is why NUMA (non-uniform memory access, i.e.
not all memory is equally fast from each processor) was invented. Linux 2.4 has some preliminary support for NUMA, but it's still in the beginning stages.
Even though there are superior alternatives out
there for most every application, the loose knit groups can't, or
won't, coordinate their movements and move at once.
oh yeah? okay, please show me a "superior" alternative to IRC, that allows people to meet on dynamically managed chatrooms (aka channels), including a system of moderation (aka ops) that works reliably (it does on IRC, that's why "l77t warriors" are reduced to generic IP-level attacks like flooding servers' connections), while being completely decentralized, using open protocols, and not requiring any form of advance registration. please show me such a system in production. if there is one, i'm interested.
Re:IRC's design is incompatable with today's inter
on
EFnet Hits Turbulence
·
· Score: 2
IRC's design has always been broken in many other ways. I remember back in 93 or so when servers would let you do a remote "/stats k" and happily send over 10k of lines back to you. Send 20 of those in a row, and you had a netsplit. I was so shocked to find that, that it didn't even occur to me that people would do such "bad things" other than once for testing. Of course, at that time packet flooding was unheard off, and a single spoofed ICMP was considered "really bad", too.
As somoene who has worked extensively on IRC server to server protocols, I can confirm what you're saying: ircd scales rather badly. It only has gone up to the current levels thanks to the massive increases in bandwith and server memory. If IRC wants to keep the model of a single network with a unique channel namespace, and a completely decentralized network of servers, then alternate routes and cyclic links become a necessity. EFnet has been going mostly in the opposite direction: a few strong central hubs, and lots of leaves. That works better than the random-spanning-tree that IRC started with.
and what I'd *really* want to see is sendmail and wu-ftpd going the way of the dodo (or just not installed by default), and replaced by postfix and openbsd-ftpd respectively.
they're going to be writing documents for political lobbying, not doing anything tech related, let alone anything to do with mozilla. i'd rather see an organization that focused on building a version of mozilla with strong privacy features, and *they* would have much better claim to the name "privacilla".
Slashdot Mirror
so does Spain! i guess idiot minds think alike too...
I've come across the opposite problem: I wrote a daemon on Linux, announced it asking for feedback especially on non-Linux systems (BSD, Solaris, etc), and got exactly none. And it was not for lack of interest; I did get a fair number of downloads. So my deamon is still in the state of "works fine under Linux, you're on your own elsewhere".
Oh well, one of these days I'll have to install FreeBSD on a spare partition...
yeah, especially in the number of bytes it takes. now, when it comes to value and interest, it's more like a word is worth half a dozen .gifs. just do a "page info" from netscape on any page on a commercial website, and go through the list of .gifs and .jpgs... most of them are completely uninformative, they're just put there to make the layout "cute" and the page slower.
yep, it's called AxKit.
space elevator? bah, I want a space escalator. imagine *that*.
I had the *exact same feeling*. Ender's Game was a good, fast and compelling read. Speaker for the Dead was an amazing broadening and deepening of both the subject and the characters. Xenocide had great quotes, great characters (though both taoism and puritanism get caricatured rather badly), great situations, and a shitty deus-ex-machina ending that fit like a soccer ball in the soup. for that reason I haven't read Children of the Mind (or whatever the following sequel is called), although I'll probably pick it up if I see it cheap enough in a 2nd hand bookstore. As for the 'shadow' series, I really don't find the premise all that compelling, nor the characters of Bean or Peter. Give me Valerie's story anytime, and I'll buy it ... no wait, no need, it's already in Speaker for the Dead.
There are *many* technologies out there to do these things right. Apache with mod_perl and either AxKit or the Templating Toolkit (or others) can do it. I hear PHP can do it using the right libraries. Java servlets, for all their verbosity, can do it using something like go.com's Tea system.
I have never used ColdFusion or ASP myself, but I know many people who have, and I've repeatedly heard that they're great for quickly building something, but very bad for managing complexity.
no comment.
just call it enamel!
be surprised then, because it really is... the word for word translation would be: "an open operating system has not only advantages".
Is it me, or did Micros~1 just admit in public that Linux has considerable advantages? I mean, how else could they be saying that it doesn't *only* have advantages?
I take it that you're saying you don't believe in Turing's thesis. could you specify a bit more what you mean by "intelligent way" as opposed to "analytical way"? specifically, do you see this 'intelligent way' as being specifiable by a good old formal-language algorithm?
The popular spots in the Himalayas *are* getting worryingly filthy. I haven't been in tha Everest area (which btw isn't that hard to access; trekking to Everest Base Camp is hugely popular, it's only going up further that gets hard), but I've been in the Annapurna range, and I was apalled at the amount of junk that litters the paths there. The camp at Thorung Phedi (last stop before the Thorung La, at 4100m or so) is quite dirty. Of course, the view from there more than compensates for it ;)
you'd think they'd pick a decent, modern and relatively clean language like Python, Pike or, a bit more esoterically, Ruby...
i have nothing against technology, in fact I love to play with it. but I'm deeply convinced that the major bottlenecks in human society and evolution are *not* technical, which means that technology isn't going to solve them, nor (I hope) to make them substantially worse. which is fine, really, as it's not its job. computers are all about easily doing things that we could already do, but were (hugely) prohibitive in cost and time. what gets branded as "innovation" from the computer world (and I don't only mean MS here) is just ridiculously simple compared to the real human problems that each of us has to grapple with, and I see no evidence at all of that changing.
seriously, most of what he says makes so much sense that it's scary. there have been documented cases of end of the world theorists since at least 2000 years ago, and people still believe it when new ones come along with new and shiny "singularity" scenarios...
Tru64 Unix may be solid and scalable, but it's not perfect either. I wouldn't be surprised if both Linux and FreeBSD had better tcp/ip these days.
that's one side of the story. the other side is that clone() doesn't support the exact POSIX threads API, and the linuxthreads library has to bend over backward (and be slow, in the process) to add this support. the situation is slowly fixing itself, but there are some human communication problems in the middle, and it takes a lot of time for a new version of linuxthreads to make it all the way to the distributions.
that *still* depends on the workload. if it's CPU-bound, it just works. if it's I/O bound, it depends on global system recourses (bus speed and bandwith), and on how well smp-threaded that particular subsystem is on the OS. one of the big improvements of the linux 2.4 kernel is precisely a major overhaul of the tcp/ip stack to smp-thread it completely. OTOH, heavy filesystem work (lots of rename(), unlink(), open(O_CREAT), etc) is not likely to scale super-well, because of filesystem locking constraints, and that's under all OSes too, which is why you generally avoid writing your big programs in such a way that they have to be messing around with lots of little files all the time (e.g that's why you use a database backend, or some kind of db format, rather than flat files, if you want a news server that scales). beyond that, in the real high end (32 procs and more), the SMP model itself starts showing its limits, which is why NUMA (non-uniform memory access, i.e. not all memory is equally fast from each processor) was invented. Linux 2.4 has some preliminary support for NUMA, but it's still in the beginning stages.
oh yeah? okay, please show me a "superior" alternative to IRC, that allows people to meet on dynamically managed chatrooms (aka channels), including a system of moderation (aka ops) that works reliably (it does on IRC, that's why "l77t warriors" are reduced to generic IP-level attacks like flooding servers' connections), while being completely decentralized, using open protocols, and not requiring any form of advance registration. please show me such a system in production. if there is one, i'm interested.
As somoene who has worked extensively on IRC server to server protocols, I can confirm what you're saying: ircd scales rather badly. It only has gone up to the current levels thanks to the massive increases in bandwith and server memory. If IRC wants to keep the model of a single network with a unique channel namespace, and a completely decentralized network of servers, then alternate routes and cyclic links become a necessity. EFnet has been going mostly in the opposite direction: a few strong central hubs, and lots of leaves. That works better than the random-spanning-tree that IRC started with.
and what I'd *really* want to see is sendmail and wu-ftpd going the way of the dodo (or just not installed by default), and replaced by postfix and openbsd-ftpd respectively.
if your users *need* comet cursor I'm very happy I don't have to be anywhere near them.
they're going to be writing documents for political lobbying, not doing anything tech related, let alone anything to do with mozilla. i'd rather see an organization that focused on building a version of mozilla with strong privacy features, and *they* would have much better claim to the name "privacilla".