I suspect that the original poster *ARGEES* with you! His modest proposal is clearly a poke at the people who actually would think of such a thing (albiet probably not in so many words) -- if you don't get this, please go study a history of satire, starting with J. Swift.
Well, this makes sense. If I take the engine out of a car it's much less likely that someone will drive off with it:) If they don't really need the extra features a Unix would give them, why not use a braindead system which happens to give the webserver complete control of the computer and has almost no capabilities to exploit?
There's a pre-alpha file manager, fsviewer. I tried it and..well..it has a ways to go, it only has the very basic funtionality and isn't very stable. It looks nice, though.:) Nothing is stopping you from writing a program with GNUStep, though. Daniel
Donst anybody get it? its a single point of communication. So that anybody can use any part of the internet. I already have that and I'm sitting in front of it. Is there *any* significant advantage from putting all these functions into one huge program over having multiple programs to do them? "Because we can" isn't a good reason:-)
I like the idea of little handheld computers for simple tasks, but what makes one program that does 500 things better (or even more integrated) than lots of programs doing one thing each?
Your reflexive flammage of anything that you perceive as a threat to yourself is exactly what I was criticizing: not Apple, not progress, not user-friendliness. If it is so plainly obvious that OSX is not a Unix, despite having a BSD UNIX core, could you please explain yourself? Insulting your listeners is not a good way to carry an argument.
Many people trash Emacs for being 'bloated'. Emacs users reply that Emacs isn't bloated, it just has 50000 elisp modules that you can load. Many people trash Mozilla for being 'bloated'. Mozilla developers reply that Mozilla isn't bloated, it just has 50000 loadable modules that you can add.
I like Emacs' "bloat". Despite the similarity, however, I am disturbed by Mozilla's bloat. Why? Let me explain... In Emacs, virtually every module I've seen, except for w3 and a few cute things like the Tower of Hanoi, is somehow related to text editing and viewing. This makes sense since Emacs' primary function for most people is as a text editor. Even the news/mailreader falls into this category, since you have to edit text to read mail and news anyway. Even these 'leverage' the abilities of Emacs to perform their operations. Mozilla, on the other hand, is a Web browser. Loadable modules that would make sense would be things like movie-playing plugins, Java VMs, interface touchups, and so on. What doesn't make sense is adding completely new functionality under the pretense of 'integration'. How is it more 'integrated' if my IRC client and Web browser are the same program? IRC clients don't need complex HTML layout; in fact, virtually the only piece of code that the two functions share is the code to transmit bytes over a network and maybe configuration infrastructure. Mail and news clients as well have *no reason* to be 'integrated' into a Web browser unless they are using HTML to send and receive messages -- and as anyone who has unexpectedly gotten one of these HTML messages can tell you, HTML email is Evil. (even with mutt, which parses mailcap and starts lynx..) FTP, on the other hand, makes some sense -- you're generally viewing structured information and retrieving files. I suspect that most of these will end up being like Emacs' w3: "My Web browser can read mail! How cute!" However, the fact that several of these projects (the mail and news clients at least) appear to be part of Mozilla's "core" distribution makes me wonder about the goals and attitudes of the project. When I heard that Mozilla was going to create a fast, clean implementation of the browser, I was ecstatic. Finally something that would just let me read Web pages! How wrong I was. I wish I had time to make my own browser, but I've already got plenty of other projects to work on. Hopefully one of the other groups working on a free browser will avoid the pitfall of overgeneralization..
Actually, I believe that GPG and PGP use a symmetric algorithm to encrypt the main body of data; the key is encrypted and stored somewhere else in the file (I guess at the beginning would make sense:) ) So if you were using a *really* insecure algorithm, someone could crack your messages one by one without ever breaking the asymmetric algorithm. As you point out though, all the symmetric cyphers are 'good enough'.
Apple = forward, quality, simplicity, elegance. unixes = backwater, cheesy, don't-change-if-ain't-broken, hacks rulz,/. rulz. Thank you for your masterful summation of the situation. Your breathtaking clarity, wit, and insight have enlivened yet another mundane day on Slashdot. Confronted with this burst of genius, we readers can only stand in awe and marvel at the enlightenment being given to us. Clearly UNIX is an obsolete piece of junk, only suitable for complete computer nerds, with no future at all: Xah Has Spoken. </sarcasm>
Daniel
PS - Apple's next operating system release *is* UNIX.
See the subject line. It lets you view info pages using Lynx-style keybindings; since I got it I actually *like* info documents, since they do have some nice features over, say, HTML, but the standard browser makes it too hard to get to them.
(as a side note, pinfo can also do manpages very nicely, including turning references into 'links')
I disagree. In fact, here at Brown Scheme is being taught as one language in an introductory CS class with no great problems. IMO it has much *less* punctuation than other programming languages (like class JavaLanguage { static void main(String args[]) {System.out.println("No this language has \"no\" punctuation at all");} } I think that either Scheme or Python will be ok; I have to admit that I don't especially like Java in general anyway, but I particularly don't think it's good for beginners. (a lot of people in the new intro class that I took last year -- which used Scheme, ML, and Java -- got through Scheme and ML just fine but fell apart trying to learn Java and all the bizarre quirks of the language, although that was partly because the teachers skimmed over Java syntax a little quicker than they should have -- OTOH this is exactly the problem with Java as an intro language. I think the only reason Java was even included in the class was the (misguided) pressure to 'teach students real-world skills')
Do you care for Open Source / Free Software or Linux? I'll take Linux...
Loyalty is an interesting thing, and the cause of much flammage, but no-one really seems to understand it. Here are some general categories and my opinions expressed briefly...
Brand/company loyalty. This I really don't get; being loyal to a company unless they're actually giving me money and a job is something that I just don't understand. Companies are out to make money, being loyal to a company or a brand is like a herd of sheep being loyal to a wolf. It doesn't work.
Product loyalty. This has more merit, although it sometimes dovetails with loyalty to a brand or company. After all, a good program is..well..a good program. However, this has a tendency to lead people to unnecessary narrow-mindedness and hosility. (look at the flamewars against ABL (Anything But Linux) that pop up here) Ok but dangerous.
Concept loyalty. That is, loyalty to a design idea or to a way of doing things. For example, someone who likes UNIX systems (raises hand) because of their particular design features, or someone who believes stridently in the GUI Jihad.
Philosophical loyalty. Loyalty to an idea above and beyond a specific implementation or method, or to a goal which is believed to be likely to Make The World a Better Place.
My feeling is that in general: -> Microsoft users are in the first category -> 'Linux advocates' tend to be in the second category -> 'Open-source' advocates tend to be in the third (I say this because despite the widespread use of the word 'philosophy', Open Source seems to really reject philosophical or ethical arguments in favor of strict utilitariansim, and even openly ridicules people who dare to contradict this approach.) -> 'Free software' advocates tend to be spread between the third and fourth.
You may have guessed my general opinions on the subject, 1<2<3<4 (although there are people strongly commited to philosophies that are absolutely ridiculous but dangerous..)
Anyway, I'm not sure what my original point was going to be, except that being committed to a single program (Linux) seems to me to be a bit parochial and you should try to broaden your sights a little. I was probably going to say something like that:)
I just don't see how Python can get, uhhh... in whack(?) again if people mess up the formatting (This is bound to happen if more than one person modifies a file).
Well, there are a couple of issues here: -> Editability. With a Python-aware editor (Emacs in python-mode for example) the indentation is handled laughably easily: Tab indents either to the current indentation or one higher, Backspace at the beginning of a line drops the indentation one level. 4-space indents (I think) so you really see the difference between the levels. No tabs at all are used AFAIK, but it doesn't matter if they are. This is really much much less of a problem than you seem to think; the only times I've ever been bitten by it were when entering complex definitions in the interpreter -- something that you rarely want to do anyway for plenty of other reasons. -> Newbie confusion. I don't really see how this (with a decent editor) is more confusing than anything else. Plus it enforces proper style >=) -> Maintainence by multiple people. Since they are all using proper editors:) the syntax shouldn't get 'out of whack' -- they will always indent 4 spaces to the left or right of the current block. However, even if they do have differing opinions on the subject, if they are at all competent they will reindent an entire block when they reindent one line. (as a matter of fact, Emacs in python-mode has a function for reindenting blocks) Python doesn't care that every block in the program isn't at the same level, it just needs indentation to be consistent within blocks. And you *could* use a program to make your indentation consistent again as long as the blocks were correct, it would just be highly specialized:) (in fact, I could probably write such a thing in an hour or two..)
In sum -- it seems to sound really bad to people when they read about it (I also went "Huh?" at first) but it is not a problem when actually using the language. (because all the people who have a problem with it got scared off I guess;-) ) I find that it actually makes things easier, since I always indent my code consistently anyway; it just means I can stop lining up braces..
If the citizenry is stupid enough to elect someone based on his haircut, so be it. It's their bed, and they'll have to sleep in it.
Unfortunately, so do the rest of us. Not that I think the solution advocated two up from this is much better. Personally, I wonder if the ideal government and country is one man on a desert island:-)
Right now voting happens on a single day, and if you can't get to the booth that day you're stuck. (unless you can cast proxy votes? but I suspect setting that up is a *huge* pain) Why not allow votes to be cast over several days? Not much help, but it would make things better for people on tight schedules. (I assume the difficulty is the trouble of keeping the voting infrastructure running for several days) More interesting is a related idea, which pretty much requires some form of computerized voting. Make the polls open for a much, much longer time -- say a year or more -- and let anyone vote at anytime, with the stipulation that the votes aren't actually counted until election day. (people who have voted would be allowed to alter their vote) The main problem here is that the running totals would have to be kept absolutely secret, and seeing how sieve-like large organizations, be they government or business, tend to be, this is a problem. Maybe they could hire some Transmeta folks:-)
Since everyone else seems to not have a sense of humor (maybe you were too close to the real thing..;) ) I thought that I'd mention that this is possibly the funniest thing I've read all week.
Hopefully someone will moderate it up, but the level of sophistication with regard to satire seems to be low already and dropping fast around here..
That the Linux kernel could be modified on-the-fly via a module is a serious security hole. That really needs to be fixed, urgently, IMHO.
Well, if you're that paranoid you can compile a kernel without module support. (you may have some difficulty with newer hardware, PnP devices and stuff though..that's a problem. Maybe sort-of modules..compile it like a module but load it into the kernel at bootup, just delay its activation and don't allow later loading of modules)
Loadable module support is always going to leave you vulnerable to this sort of thing, though.
First I'd like to correct the stereotype if I may:) The stereotype that I've seen is that 'geeks' are arrogant and obnoxious libertarian neo-Social-Darwinists interested primarily in making money. From my own observation I'd say that a significant number of so-called 'geek's fit fairly well into this category. That said -- "They are liberal...they do not believe in God..." This is one of the more common examples of precisely what you are complaining about! I know a large number of extremely religious liberals; to assume that one must be conservative to be religious (or vice versa) is a pretty good example of either your own prejudices or the inadequacy of convenient categories to properly capture the range of possible human beliefs. I prefer the second option. (and I may have done a little of this myself in this very post -- it's a human trait to try to categorize the uncategorizable. That doesn't mean that it's any more accurate, but it always happens.)
You even mention HURD later. What's wrong with it? (aside from the fact that it's unfinished) It's free software (and of course 'OSS') and much more modern than Linux (now I'm sure someone's going to complain that it is 'less modern' than the latest greatest research operating system..research OSes are good but it takes a while for the concepts to percolate to 'real systems') -- in fact, I believe its architecture may be similar to BeOS from what little I know of Be.
I usually try to avoid getting into distribution discussions--Debian users have an (unfortunately) deserved reputation for being fanatical and I don't like to enhance it. BUT:
Actually, I'm not sure that these people will leave Debian for a long time. Debian is trying, slowly but surely, to evolve into a general operating system where the kernel is just another option. The most radical example is, of course, Debian GNU/Hurd, but there has also been talk of making a Debian GNU/BSD and of course Debian has ports to all sorts of weird hardware:) This is the most interesting thing (to me) about it -- a configuration (we really are making a central config tool, we're just doing it right), package management, and application system that is mostly agnostic as to what it's running on.
So, I sit down to try and port some of these things to the OS I'm currently on. Uh oh. It uses inline assembler, completely uncommented of course. Or, its completely reliant on the bastardized Linux kernel headers. Or it needs a "convenience" device that exists on Linux despite the fact that the stuff could be done completely fine at the application level.
I think you're exaggerating the problem a little; Debian has been running into problems porting stuff en masse to the Hurd but only a few really bad cases. Not to say that this isn't a problem -- in fact it may be worse than you think; I was told once on/. by a programmer that he deliberately wrote programs that would not run on other operating systems (Hurd was the specific subject, actually -- I assume he was using/proc heavily)
Slashdot Mirror
I--R--O--N--Y. S--A--T--I--R--E.
I suspect that the original poster *ARGEES* with you! His modest proposal is clearly a poke at the people who actually would think of such a thing (albiet probably not in so many words) -- if you don't get this, please go study a history of satire, starting with J. Swift.
Daniel
Well, this makes sense. If I take the engine out of a car it's much less likely that someone will drive off with it :)
If they don't really need the extra features a Unix would give them, why not use a braindead system which happens to give the webserver complete control of the computer and has almost no capabilities to exploit?
Daniel
They might read it, but I'm not sure about the polish.. :)
Daniel
Mutt has extremely good support for pgp, pgp5, and gpg.
Daniel
There's a pre-alpha file manager, fsviewer. I tried it and..well..it has a ways to go, it only has the very basic funtionality and isn't very stable. It looks nice, though. :)
Nothing is stopping you from writing a program with GNUStep, though.
Daniel
Donst anybody get it? its a single point of communication. So that anybody can use any part of the internet. :-)
I already have that and I'm sitting in front of it. Is there *any* significant advantage from putting all these functions into one huge program over having multiple programs to do them? "Because we can" isn't a good reason
I like the idea of little handheld computers for simple tasks, but what makes one program that does 500 things better (or even more integrated) than lots of programs doing one thing each?
Daniel
Your reflexive flammage of anything that you perceive as a threat to yourself is exactly what I was criticizing: not Apple, not progress, not user-friendliness. If it is so plainly obvious that OSX is not a Unix, despite having a BSD UNIX core, could you please explain yourself? Insulting your listeners is not a good way to carry an argument.
Daniel
Many people trash Emacs for being 'bloated'. Emacs users reply that Emacs isn't bloated, it just has 50000 elisp modules that you can load.
Many people trash Mozilla for being 'bloated'. Mozilla developers reply that Mozilla isn't bloated, it just has 50000 loadable modules that you can add.
I like Emacs' "bloat". Despite the similarity, however, I am disturbed by Mozilla's bloat. Why? Let me explain...
In Emacs, virtually every module I've seen, except for w3 and a few cute things like the Tower of Hanoi, is somehow related to text editing and viewing. This makes sense since Emacs' primary function for most people is as a text editor. Even the news/mailreader falls into this category, since you have to edit text to read mail and news anyway. Even these 'leverage' the abilities of Emacs to perform their operations.
Mozilla, on the other hand, is a Web browser. Loadable modules that would make sense would be things like movie-playing plugins, Java VMs, interface touchups, and so on. What doesn't make sense is adding completely new functionality under the pretense of 'integration'. How is it more 'integrated' if my IRC client and Web browser are the same program? IRC clients don't need complex HTML layout; in fact, virtually the only piece of code that the two functions share is the code to transmit bytes over a network and maybe configuration infrastructure. Mail and news clients as well have *no reason* to be 'integrated' into a Web browser unless they are using HTML to send and receive messages -- and as anyone who has unexpectedly gotten one of these HTML messages can tell you, HTML email is Evil. (even with mutt, which parses mailcap and starts lynx..) FTP, on the other hand, makes some sense -- you're generally viewing structured information and retrieving files.
I suspect that most of these will end up being like Emacs' w3: "My Web browser can read mail! How cute!" However, the fact that several of these projects (the mail and news clients at least) appear to be part of Mozilla's "core" distribution makes me wonder about the goals and attitudes of the project.
When I heard that Mozilla was going to create a fast, clean implementation of the browser, I was ecstatic. Finally something that would just let me read Web pages! How wrong I was. I wish I had time to make my own browser, but I've already got plenty of other projects to work on. Hopefully one of the other groups working on a free browser will avoid the pitfall of overgeneralization..
Daniel
Actually, I believe that GPG and PGP use a symmetric algorithm to encrypt the main body of data; the key is encrypted and stored somewhere else in the file (I guess at the beginning would make sense :) ) So if you were using a *really* insecure algorithm, someone could crack your messages one by one without ever breaking the asymmetric algorithm.
As you point out though, all the symmetric cyphers are 'good enough'.
Daniel
I think that 'appropriate/inappropriate' would be better and more accurate.
Daniel
Apple = forward, quality, simplicity, elegance. /. rulz.
unixes = backwater, cheesy, don't-change-if-ain't-broken, hacks rulz,
Thank you for your masterful summation of the situation. Your breathtaking clarity, wit, and insight have enlivened yet another mundane day on Slashdot. Confronted with this burst of genius, we readers can only stand in awe and marvel at the enlightenment being given to us. Clearly UNIX is an obsolete piece of junk, only suitable for complete computer nerds, with no future at all: Xah Has Spoken.
</sarcasm>
Daniel
PS - Apple's next operating system release *is* UNIX.
See the subject line. It lets you view info pages using Lynx-style keybindings; since I got it I actually *like* info documents, since they do have some nice features over, say, HTML, but the standard browser makes it too hard to get to them.
(as a side note, pinfo can also do manpages very nicely, including turning references into 'links')
Daniel
I disagree. In fact, here at Brown Scheme is being taught as one language in an introductory CS class with no great problems. IMO it has much *less* punctuation than other programming languages (like class JavaLanguage { static void main(String args[]) {System.out.println("No this language has \"no\" punctuation at all");} }
I think that either Scheme or Python will be ok; I have to admit that I don't especially like Java in general anyway, but I particularly don't think it's good for beginners. (a lot of people in the new intro class that I took last year -- which used Scheme, ML, and Java -- got through Scheme and ML just fine but fell apart trying to learn Java and all the bizarre quirks of the language, although that was partly because the teachers skimmed over Java syntax a little quicker than they should have -- OTOH this is exactly the problem with Java as an intro language. I think the only reason Java was even included in the class was the (misguided) pressure to 'teach students real-world skills')
Daniel
Loyalty is an interesting thing, and the cause of much flammage, but no-one really seems to understand it. Here are some general categories and my opinions expressed briefly...
My feeling is that in general:
-> Microsoft users are in the first category
-> 'Linux advocates' tend to be in the second category
-> 'Open-source' advocates tend to be in the third (I say this because despite the widespread use of the word 'philosophy', Open Source seems to really reject philosophical or ethical arguments in favor of strict utilitariansim, and even openly ridicules people who dare to contradict this approach.)
-> 'Free software' advocates tend to be spread between the third and fourth.
You may have guessed my general opinions on the subject, 1<2<3<4 (although there are people strongly commited to philosophies that are absolutely ridiculous but dangerous..)
Anyway, I'm not sure what my original point was going to be, except that being committed to a single program (Linux) seems to me to be a bit parochial and you should try to broaden your sights a little. I was probably going to say something like that
Daniel
People are allowed to disagree with a company's actions even if it spent $450 million dollars on them.
Daniel
I just don't see how Python can get, uhhh... in whack(?) again if people mess up the formatting (This is bound to happen if more than one person modifies a file).
:) the syntax shouldn't get 'out of whack' -- they will always indent 4 spaces to the left or right of the current block. However, even if they do have differing opinions on the subject, if they are at all competent they will reindent an entire block when they reindent one line. (as a matter of fact, Emacs in python-mode has a function for reindenting blocks) Python doesn't care that every block in the program isn't at the same level, it just needs indentation to be consistent within blocks. And you *could* use a program to make your indentation consistent again as long as the blocks were correct, it would just be highly specialized :) (in fact, I could probably write such a thing in an hour or two..)
;-) ) I find that it actually makes things easier, since I always indent my code consistently anyway; it just means I can stop lining up braces..
Well, there are a couple of issues here:
-> Editability. With a Python-aware editor (Emacs in python-mode for example) the indentation is handled laughably easily: Tab indents either to the current indentation or one higher, Backspace at the beginning of a line drops the indentation one level. 4-space indents (I think) so you really see the difference between the levels. No tabs at all are used AFAIK, but it doesn't matter if they are. This is really much much less of a problem than you seem to think; the only times I've ever been bitten by it were when entering complex definitions in the interpreter -- something that you rarely want to do anyway for plenty of other reasons.
-> Newbie confusion. I don't really see how this (with a decent editor) is more confusing than anything else. Plus it enforces proper style >=)
-> Maintainence by multiple people. Since they are all using proper editors
In sum -- it seems to sound really bad to people when they read about it (I also went "Huh?" at first) but it is not a problem when actually using the language. (because all the people who have a problem with it got scared off I guess
Daniel
If the citizenry is stupid enough to elect someone based on his haircut, so be it. It's their bed, and they'll have to sleep in it.
:-)
Unfortunately, so do the rest of us. Not that I think the solution advocated two up from this is much better. Personally, I wonder if the ideal government and country is one man on a desert island
Daniel
Right now voting happens on a single day, and if you can't get to the booth that day you're stuck. (unless you can cast proxy votes? but I suspect setting that up is a *huge* pain) Why not allow votes to be cast over several days? Not much help, but it would make things better for people on tight schedules. (I assume the difficulty is the trouble of keeping the voting infrastructure running for several days) :-)
More interesting is a related idea, which pretty much requires some form of computerized voting. Make the polls open for a much, much longer time -- say a year or more -- and let anyone vote at anytime, with the stipulation that the votes aren't actually counted until election day. (people who have voted would be allowed to alter their vote) The main problem here is that the running totals would have to be kept absolutely secret, and seeing how sieve-like large organizations, be they government or business, tend to be, this is a problem. Maybe they could hire some Transmeta folks
Daniel
Since everyone else seems to not have a sense of humor (maybe you were too close to the real thing.. ;) ) I thought that I'd mention that this is possibly the funniest thing I've read all week.
Hopefully someone will moderate it up, but the level of sophistication with regard to satire seems to be low already and dropping fast around here..
Daniel
That the Linux kernel could be modified on-the-fly via a module is a serious security hole. That really needs to be fixed, urgently, IMHO.
Well, if you're that paranoid you can compile a kernel without module support. (you may have some difficulty with newer hardware, PnP devices and stuff though..that's a problem. Maybe sort-of modules..compile it like a module but load it into the kernel at bootup, just delay its activation and don't allow later loading of modules)
Loadable module support is always going to leave you vulnerable to this sort of thing, though.
Daniel
First I'd like to correct the stereotype if I may :) The stereotype that I've seen is that 'geeks' are arrogant and obnoxious libertarian neo-Social-Darwinists interested primarily in making money. From my own observation I'd say that a significant number of so-called 'geek's fit fairly well into this category. That said --
"They are liberal...they do not believe in God..." This is one of the more common examples of precisely what you are complaining about! I know a large number of extremely religious liberals; to assume that one must be conservative to be religious (or vice versa) is a pretty good example of either your own prejudices or the inadequacy of convenient categories to properly capture the range of possible human beliefs. I prefer the second option. (and I may have done a little of this myself in this very post -- it's a human trait to try to categorize the uncategorizable. That doesn't mean that it's any more accurate, but it always happens.)
Daniel
palantir:~> dlocate -s ssh
:)
Package: ssh
Status: install ok installed
Priority: optional
Section: non-US/non-free
Case closed
Daniel
You even mention HURD later. What's wrong with it? (aside from the fact that it's unfinished) It's free software (and of course 'OSS') and much more modern than Linux (now I'm sure someone's going to complain that it is 'less modern' than the latest greatest research operating system..research OSes are good but it takes a while for the concepts to percolate to 'real systems') -- in fact, I believe its architecture may be similar to BeOS from what little I know of Be.
Daniel
I usually try to avoid getting into distribution discussions--Debian users have an (unfortunately) deserved reputation for being fanatical and I don't like to enhance it. BUT:
:) This is the most interesting thing (to me) about it -- a configuration (we really are making a central config tool, we're just doing it right), package management, and application system that is mostly agnostic as to what it's running on.
Actually, I'm not sure that these people will leave Debian for a long time. Debian is trying, slowly but surely, to evolve into a general operating system where the kernel is just another option. The most radical example is, of course, Debian GNU/Hurd, but there has also been talk of making a Debian GNU/BSD and of course Debian has ports to all sorts of weird hardware
Daniel
So, I sit down to try and port some of these things to the OS I'm currently on. Uh oh. It uses inline assembler, completely uncommented of course. Or, its completely reliant on the bastardized Linux kernel headers. Or it needs a "convenience" device that exists on Linux despite the fact that the stuff could be done completely fine at the application level.
/. by a programmer that he deliberately wrote programs that would not run on other operating systems (Hurd was the specific subject, actually -- I assume he was using /proc heavily)
I think you're exaggerating the problem a little; Debian has been running into problems porting stuff en masse to the Hurd but only a few really bad cases. Not to say that this isn't a problem -- in fact it may be worse than you think; I was told once on
Daniel