Re:dan bernstein's position on this
on
DNSSEC: Good Enough?
·
· Score: 4, Interesting
djb's points about dnssec seem reasonable, but his proposed solution `nym' seems quite nutty.
He basically proposes only allowing a form of hostname which is (1) too long to type manually, and (2) includes long random-looking strings. His justification for this is `users seem to do alright with bookmarks, and as soon as everything is links, no problem!'
Is he living on the same earth we do? It's going to be a long time before manually enterable -- and verifiable -- hostnames become redundant (if they ever do).
Um, sure, but the exact same thing applies to imake -- and unless you're using a very common system with a bog-standard installation, imake fails more often than it succeeds.
Imake and autoconf take fundamentally different approaches to configuration: imake says `give me a system type I know about, and I'll give you some makefiles' -- great, unless of course it doesn't know about your system, or if you've changed something. Autoconf, on the other hand says `I'll grovel around on your system and see what things it implements that I know about, and try to come up with a configuration' -- this works much better on disparate (though not so disparate that the probing just fails, or the set of features implemented is simply outside of what autoconf can use) or hacked platforms.
The imake approach is great if there's only a fairly small number of system types in use, but fails miserably when this isn't true.
In the early days of X, imake's approach was probably reasonable, but as X was ported to more and more systems, the number of unix variants exploded, and the number of home-grown linux variants &c increased (which all had different library versions installed), it became a bad joke. Of course this was hard for automake too, but it was far better place to cope with such a challenge.
Now, as many proprietary unix variants fall by the wayside, and the interfaces used in typical linux/bsd systems have become a bit more stable, maybe imake has become more practical again (I don't really know).
Another issue entirely is the implementations. I've wrestled with both, and as far as I'm concerned, both suck. No points there.:-):-(
Personally I think that the autoconf method is far more elegant; in my mind the best solution would be something like autoconf, except rewritten in a language that wasn't m4, and perhaps more support for external input to tests -- for instance, if you could override the method it uses to check for library functions to always call some system-defined checker instead, it could be made more portable even to systems on which its (often rather shaky) testing methods don't work.
I expect so, although I don't know off hand his education.
I believe he has a BSc in physics, from Harvard (though something like 25 years ago). He tried hard not to graduate, but they eventually forced him to...
Sco's stock price drops below 11 and their claims grow louder and more rediculous. My only question is how do they think they will avoid the inevitable insider trading and fraud investigations.
My question is: What are they going to do when their stock price hits, oh say, 1?
It's going to take some serious talent (and a lot of LSD) to get that freaky...
Ever get an email from Yahoo? Its ALL HTML. There is no plain text. Yahoo is bigger than hotmail, I am sure somebody has sent to you from there.
Sure, I regularly get mail from yahoo, and not a drop of html* (this is from non-techie types too). I have no idea what yahoo's defaults are, but they don't seem to be a problem for me.
Nintendo is not run by the Keebler Elves. They pull the same shit as everyone else.
Huh? Were you asleep during the last round of hype wars?
While Sony and MS were trying to outdo each other in making ludicrous hyper-inflated performance claims for their respective hardware, Nintendo gave far lower numbers -- which they said were `realistic.' Now that the hardware and games are available to compare for real, it's clear that the gamecube is far more powerful than the PS2, and only slightly less powerful than the xbox.
So, yeah, Nintendo appears to be run by elves. Kinda appropriate actually...:-)
I've filtered pure html mail for a long time -- it's a highly effective way to get rid of spam, and nobody I know (even my non-computer-adept relatives) is so clueless as to send pure-html email.
I've noticed recently that spammers are trying to get around this, not by putting their spam in text form, but by trying to disguise the html using multipart etc (it's still easy to automatically identify though).
I think the point is that emacs isn't a tool, it's the whole toolset. If you're an emacs user, you write little lisp functions and macros to do everything that a unix user might use pipelines to do.
Of course an emacs user can also use unix tools as part of his toolset, because emacs is fairly deft with processes. The reverse isn't as easy, which is why some people flame endlessly about emacs being `inappropriate' -- but honestly, real emacs users don't care; they've got the toolkit they need.
I don't submit patches to Microsoft. Why should I submit patches to some open source developer?
Because that will get you what you want, faster than if you don't submit patches.
Microsoft implements features because it feels that X percentage of users will like them, and those users will pony up lots of money to get them -- even if, in fact, the features aren't really what each user really wants; the users basically take what they can get.
If a particular user wants a particular feature in a microsoft product, does he have the option of implementing it himself or paying someone to do so? HA ha -- no! He's fucked.
So there you have it: if your needs are precisely aligned with those of the masses of users, and you never ever want to add your own code, or fix a small bug, or deviate in the least from the norm, well, maybe microsoft is your kind of company. Go for it, send them your money and relax.
It's supremely hypocritical to argue that open source turns out better code for people to use
Huh? I never said that. I like free software because -- get this -- it's free (as in I have the ability to change it myself).
Arguments about the technical quality of free software are interesting, but it obviously depends highly on there being a sufficient quantity of interested developers (so some FS is great, and other FS sucks), and are ultimately rather beside the point.
If open source is simply a little playground for developers, then why should anyone else care?
No one said you have to care. But -- unlike with microsoft's software -- you're allowed to care, and that's what makes all the difference.
That's fine as a shortcut. That sucks as interface as your primary interface. Why? Because it violates two principles of UI: Space and visibility.
All primary tools should be readily visible, and should be in a specific locale of space where the user can reliably remember where they are and locate them again in the future.
"Space and visibility" might be reasonable for novice user-interfaces (though it's often overused to the extent that it does more damage than good), but cramming everything into toolbars is a rotten way to make a user-interface for experts -- and it's the experts that you're claiming the gimp is no good for.
I gotta hope you're a sysadmin for a non-technical company, where that attitude makes a slight bit of sense.
Of course I don't randomly install a new OS at work (at least on a permanent, networked machine) without giving the sysadmins here a heads-up, because that would be rude, and I want to work with them, not against them. But when it comes down to it, the employees here know what they're doing, and stupid little control-freak anti-user crap like `locked-down boot media' slows me down. Don't do that.
Sorry, this is turning into a bit of a rant, but this sort of control-freak mentality is a big problem at big companies. Where I work the local machines are not an issue because the sysadmins are local too; I can talk to them, and they're reasonable people. But for instance, if I want to resolve an issue with the company firewall (`I need to use protocol X, I don't think it's a security issue, please let it through'), they more often than not simply try to ignore the problem, bury their heads in the sand, and pretend that everything's OK. If they talked to me, and raised substantive objections to my request, I'd understand -- but they don't; every single time, it's fingers in the ears and La La La La until they think everybody's gone.
> the Galeon developers do not seem to be 100% behind GNOME's goals
Yeah, there's something vaguely creepy about the above quoted sentence... It smacks way to much of petty corporate (or academic!) politics.
I really doubt everybody in the Gnome project is in such unanimous agreement about every single one of their goals -- and that's good, because some degree of dissension is healthy!
Oh, I don't know; it strongly depends on the person, anyway. I'm constantly meeting Japanese who say they haven't studied English since middle-school, but who definitely still can speak it to a functional level -- sure it's slow and rusty, but man, they haven't studied in 15 years! Just the sheer amount of vocabulary they remember boggles my mind (about all I can remember from my high-school french is `Je ne sais pas').
I know the Japanese english-education system is widely derided, and I'm sure many people don't learn much, but it's really amazing what a bit of individual dedication can do...
What I do find odd is that there doesn't really seem to be all that much difference between the above-mentioned sorts, and the ones who are actively studying english as adults.
Perhaps the biggest of all: no changesets, you have to check in changed files one by one, and there's no association between them.
If someone else happens to update from the repository while you're in the middle of doing this, well, maybe they're screwed, maybe they're not, cross your fingers! I've spent lots of time while checking big changes into CVS figuring out what the `least damaging' order is...
[Of course there's also the lack of support for distributed repositories, but I guess that's more rare than changesets in other systems too.]
Don't mean to reply to myself, but I forgot to add that most of the replies in the LKML are favorable to BK and Larry.
That has little to do with the content of the thread. The LKML community long ago split into `Bitkeeper is evil' and `I don't want to know, just let me use it' camps, and if you read the thread, it's the same old people bitching at one another.
Once again, ignorance of what Windows can do is no excuse.
Er, I thought that was the whole point of windows -- that you could use it easily despite being kinda ignorant. If you need to rely on command-line interfaces and configuration files anyway, then why not do it properly and use linux/unix in the first place?
This is due to the fact that the original author was australian, and `disc' is apparently OK there. Of course it's still stupid because `disk' is far more common in linux (the kernel and otherwise), and consistency in user-visible naming should trump the author's local conventions...
devfs is also similar to what FreeBSD has had for years. Dynamic device files make sense, there is no way around it.
I think most people agree it's a nice idea in theory, but the majority of kernel developers seem to hate the actual devfs implementation. There are several who have been working on ways to completely replace it with something more lightweight and elegant.
I'm somewhat proud that I got a 0% match for Lyndon LaRouche
Me too!
What I'm curious about is whether there's anyone who didn't get 0% for L.L. -- I'm fairly liberal, you're libertarian, and I see other responses to this story by conservatives who also got 0% for L.L.
It would be oddly impressive if LaRouche somehow manages to be 100% in disagreement with everybody else on earth... and yet still runs for political office!
Yeah, but `good' space opera manages to have all that stuff (bad acting, science, etc), but still be compelling, either because it's exciting or strikes a chord with man's yearning for the unknown or something.
B.G., on the other hand, was just sort of... blah. It was like space opera written by your accountant.
Slashdot Mirror
djb's points about dnssec seem reasonable, but his proposed solution `nym' seems quite nutty.
He basically proposes only allowing a form of hostname which is (1) too long to type manually, and (2) includes long random-looking strings. His justification for this is `users seem to do alright with bookmarks, and as soon as everything is links, no problem!'
Is he living on the same earth we do? It's going to be a long time before manually enterable -- and verifiable -- hostnames become redundant (if they ever do).
829,000 lines for symmetrical multiprocessing code in Linux? I don't have the stats to hand but I seriously doubt that.
:-)
Indeed; it's an absurd number, I'd really like to know how they claim to have derived it!
In fact, here are some quick line counts (courtesy of `wc -l') for linux-2.6.0-test3:
kernel/ -- 25,951
mm/ -- 21,635
arch/i386/ -- 70,289
Those are the main places I can think of that might contain SMP-specific code.
The vast majority of linux is in drivers, but they seem to not care about those (some other big subdirs: fs/ -- 510,786 and net/ -- 323,152).
Maybe they're counting every line that takes or releases a spin-lock as being `their idea'...
Um, sure, but the exact same thing applies to imake -- and unless you're using a very common system with a bog-standard installation, imake fails more often than it succeeds.
:-) :-(
Imake and autoconf take fundamentally different approaches to configuration: imake says `give me a system type I know about, and I'll give you some makefiles' -- great, unless of course it doesn't know about your system, or if you've changed something. Autoconf, on the other hand says `I'll grovel around on your system and see what things it implements that I know about, and try to come up with a configuration' -- this works much better on disparate (though not so disparate that the probing just fails, or the set of features implemented is simply outside of what autoconf can use) or hacked platforms.
The imake approach is great if there's only a fairly small number of system types in use, but fails miserably when this isn't true.
In the early days of X, imake's approach was probably reasonable, but as X was ported to more and more systems, the number of unix variants exploded, and the number of home-grown linux variants &c increased (which all had different library versions installed), it became a bad joke. Of course this was hard for automake too, but it was far better place to cope with such a challenge.
Now, as many proprietary unix variants fall by the wayside, and the interfaces used in typical linux/bsd systems have become a bit more stable, maybe imake has become more practical again (I don't really know).
Another issue entirely is the implementations. I've wrestled with both, and as far as I'm concerned, both suck. No points there.
Personally I think that the autoconf method is far more elegant; in my mind the best solution would be something like autoconf, except rewritten in a language that wasn't m4, and perhaps more support for external input to tests -- for instance, if you could override the method it uses to check for library functions to always call some system-defined checker instead, it could be made more portable even to systems on which its (often rather shaky) testing methods don't work.
I believe he has a BSc in physics, from Harvard (though something like 25 years ago). He tried hard not to graduate, but they eventually forced him to...
Sco's stock price drops below 11 and their claims grow louder and more rediculous. My only question is how do they think they will avoid the inevitable insider trading and fraud investigations.
My question is: What are they going to do when their stock price hits, oh say, 1?
It's going to take some serious talent (and a lot of LSD) to get that freaky...
Ever get an email from Yahoo? Its ALL HTML. There is no plain text. Yahoo is bigger than hotmail, I am sure somebody has sent to you from there.
:-)
Sure, I regularly get mail from yahoo, and not a drop of html* (this is from non-techie types too). I have no idea what yahoo's defaults are, but they don't seem to be a problem for me.
* Excepting, of course, spam...
Nintendo is not run by the Keebler Elves. They pull the same shit as everyone else.
:-)
Huh? Were you asleep during the last round of hype wars?
While Sony and MS were trying to outdo each other in making ludicrous hyper-inflated performance claims for their respective hardware, Nintendo gave far lower numbers -- which they said were `realistic.' Now that the hardware and games are available to compare for real, it's clear that the gamecube is far more powerful than the PS2, and only slightly less powerful than the xbox.
So, yeah, Nintendo appears to be run by elves. Kinda appropriate actually...
Really, you got booed??? That's a bit bizarre...
I've filtered pure html mail for a long time -- it's a highly effective way to get rid of spam, and nobody I know (even my non-computer-adept relatives) is so clueless as to send pure-html email.
I've noticed recently that spammers are trying to get around this, not by putting their spam in text form, but by trying to disguise the html using multipart etc (it's still easy to automatically identify though).
I think the point is that emacs isn't a tool, it's the whole toolset. If you're an emacs user, you write little lisp functions and macros to do everything that a unix user might use pipelines to do.
Of course an emacs user can also use unix tools as part of his toolset, because emacs is fairly deft with processes. The reverse isn't as easy, which is why some people flame endlessly about emacs being `inappropriate' -- but honestly, real emacs users don't care; they've got the toolkit they need.
Since when was Perl `small' (or for that matter `beautiful')?
Really Perl's much closer to emacs than to grep...
I don't submit patches to Microsoft. Why should I submit patches to some open source developer?
Because that will get you what you want, faster than if you don't submit patches.
Microsoft implements features because it feels that X percentage of users will like them, and those users will pony up lots of money to get them -- even if, in fact, the features aren't really what each user really wants; the users basically take what they can get.
If a particular user wants a particular feature in a microsoft product, does he have the option of implementing it himself or paying someone to do so? HA ha -- no! He's fucked.
So there you have it: if your needs are precisely aligned with those of the masses of users, and you never ever want to add your own code, or fix a small bug, or deviate in the least from the norm, well, maybe microsoft is your kind of company. Go for it, send them your money and relax.
It's supremely hypocritical to argue that open source turns out better code for people to use
Huh? I never said that. I like free software because -- get this -- it's free (as in I have the ability to change it myself).
Arguments about the technical quality of free software are interesting, but it obviously depends highly on there being a sufficient quantity of interested developers (so some FS is great, and other FS sucks), and are ultimately rather beside the point.
If open source is simply a little playground for developers, then why should anyone else care?
No one said you have to care. But -- unlike with microsoft's software -- you're allowed to care, and that's what makes all the difference.
That wasn't what was said in the post. Apparently I've offended a few GPL fans who are ignoring political process and litigation dynamics.
Um, I hate to break it to you, but your original post reads like an anti-GPL rant written by a conspiracy freak.
Maybe you had a valid point in there somewhere, but damned if I can find it...
Feel free to submit patches.
That's fine as a shortcut. That sucks as interface as your primary interface. Why? Because it violates two principles of UI: Space and visibility.
All primary tools should be readily visible, and should be in a specific locale of space where the user can reliably remember where they are and locate them again in the future.
"Space and visibility" might be reasonable for novice user-interfaces (though it's often overused to the extent that it does more damage than good), but cramming everything into toolbars is a rotten way to make a user-interface for experts -- and it's the experts that you're claiming the gimp is no good for.
So are they experts or not?
I gotta hope you're a sysadmin for a non-technical company, where that attitude makes a slight bit of sense.
Of course I don't randomly install a new OS at work (at least on a permanent, networked machine) without giving the sysadmins here a heads-up, because that would be rude, and I want to work with them, not against them. But when it comes down to it, the employees here know what they're doing, and stupid little control-freak anti-user crap like `locked-down boot media' slows me down. Don't do that.
Sorry, this is turning into a bit of a rant, but this sort of control-freak mentality is a big problem at big companies. Where I work the local machines are not an issue because the sysadmins are local too; I can talk to them, and they're reasonable people. But for instance, if I want to resolve an issue with the company firewall (`I need to use protocol X, I don't think it's a security issue, please let it through'), they more often than not simply try to ignore the problem, bury their heads in the sand, and pretend that everything's OK. If they talked to me, and raised substantive objections to my request, I'd understand -- but they don't; every single time, it's fingers in the ears and La La La La until they think everybody's gone.
> the Galeon developers do not seem to be 100% behind GNOME's goals
Yeah, there's something vaguely creepy about the above quoted sentence... It smacks way to much of petty corporate (or academic!) politics.
I really doubt everybody in the Gnome project is in such unanimous agreement about every single one of their goals -- and that's good, because some degree of dissension is healthy!
Oh, I don't know; it strongly depends on the person, anyway. I'm constantly meeting Japanese who say they haven't studied English since middle-school, but who definitely still can speak it to a functional level -- sure it's slow and rusty, but man, they haven't studied in 15 years! Just the sheer amount of vocabulary they remember boggles my mind (about all I can remember from my high-school french is `Je ne sais pas').
I know the Japanese english-education system is widely derided, and I'm sure many people don't learn much, but it's really amazing what a bit of individual dedication can do...
What I do find odd is that there doesn't really seem to be all that much difference between the above-mentioned sorts, and the ones who are actively studying english as adults.
Did I miss any?
Perhaps the biggest of all: no changesets, you have to check in changed files one by one, and there's no association between them.
If someone else happens to update from the repository while you're in the middle of doing this, well, maybe they're screwed, maybe they're not, cross your fingers! I've spent lots of time while checking big changes into CVS figuring out what the `least damaging' order is...
[Of course there's also the lack of support for distributed repositories, but I guess that's more rare than changesets in other systems too.]
Don't mean to reply to myself, but I forgot to add that most of the replies in the LKML are favorable to BK and Larry.
That has little to do with the content of the thread. The LKML community long ago split into `Bitkeeper is evil' and `I don't want to know, just let me use it' camps, and if you read the thread, it's the same old people bitching at one another.
Once again, ignorance of what Windows can do is no excuse.
Er, I thought that was the whole point of windows -- that you could use it easily despite being kinda ignorant. If you need to rely on command-line interfaces and configuration files anyway, then why not do it properly and use linux/unix in the first place?
"I can't think of anything meaningful to say."
This is due to the fact that the original author was australian, and `disc' is apparently OK there. Of course it's still stupid because `disk' is far more common in linux (the kernel and otherwise), and consistency in user-visible naming should trump the author's local conventions...
devfs is also similar to what FreeBSD has had for years. Dynamic device files make sense, there is no way around it.
I think most people agree it's a nice idea in theory, but the majority of kernel developers seem to hate the actual devfs implementation. There are several who have been working on ways to completely replace it with something more lightweight and elegant.
I'm somewhat proud that I got a 0% match for Lyndon LaRouche
Me too!
What I'm curious about is whether there's anyone who didn't get 0% for L.L. -- I'm fairly liberal, you're libertarian, and I see other responses to this story by conservatives who also got 0% for L.L.
It would be oddly impressive if LaRouche somehow manages to be 100% in disagreement with everybody else on earth... and yet still runs for political office!
Yeah, but `good' space opera manages to have all that stuff (bad acting, science, etc), but still be compelling, either because it's exciting or strikes a chord with man's yearning for the unknown or something.
... blah. It was like space opera written by your accountant.
B.G., on the other hand, was just sort of
[of course as a kid, I watched it anyway!]