Ego doesn't bother me. I've met a fairly large number of people with egos who entered the room long before they did, and I can say with some certainty that such people tend to be no more or less correct than you would find in the general population.
It is very important to remember that Microsoft (maker of Windows(tm) operating system) is a very different beast from Microsoft Research. MSR is a true research organization, and assuming that work done there will in any way translate into Microssoft products is a long-shot at best. Probably the most likely outcome of this kind of work is that some of the technology will be fed back into parts of various MS products.
I would think that if any part of his book speculated that such efficiency would be derived from CA, then that would be a reasonable request. That's not what the book is about. Give it a read.
I might disagree with some of it, but that's not the point. You've issued some very reasonable concerns about the BOOK and its CONTENT. That's fine.
What I was upset about were ad hominem attacks agaist Wolfram of the form "it wasn't peer reviewed, so he's a [kook,wacko,asshole,etc]" (yeah, someone actually called him an asshole because he was too egotistical and didn't accept peer review, though it was in another thread in this article).
I don't agree with his conclusions. I think that ultimately they will be shown to be a bit too simplistic, even if they're partially accurate. But, I refuse to write the man off as mentally abberant just because he a) wrote a book and b) didn't find every single reference to every other person who has covered the same ground before writing it. It's NOT an accedmic paper, and should not be treated as such. It's a very good book about a particular field of mathematics and an interesting treatment of one man's view of their impact on metaphysics.
"You can't get taken seriously as an academic unless others get the chance to review your work"
That's just wrong, and only true of extremists.
What is and SHOULD be true is that if your work is not peer reviewed it is not accepted as mainstream science unless it is borne out by some external source later (Fermat's Last Theorem comes to mind). That is, you shouldn't go citing it as an authoritative source of anything, nor should people feel the need to defend new work on the basis of information found therein.
But to say that a PERSON cannot be taken seriously because they wrote a book is just silly. Was he a "kook" and/or "mocked" before he wrote a book? This guy isn't a university researcher out looking for handouts from the government. He's just a guy with some good ideas (and some far-fetched conclusions that may or may not have any bearing on reality). To call him a kook for publishing a book on these topics is the worst kind of elitism, and outright damaging to those who might follow in his footsteps and share what they have to say.
I'm a child of the information age (barely, I'm almost too old for it), and my feeling is that everyone should share the bounty of their thoughts. If you want to fit into some particular community, fine, you respect that community's norms, but if you want to share, go for it.
This guy, far from being a kook, is actually quite intellegent and rational. Yes, he makes some bold stabs in his conclusions, but that's not kooky, it's just intellectually risky. I don't see anything in that book that claims he's sure he's right or where he gives the written equivalent of the finger to anyone who disagrees.
The impression that I got was that there were a few relatively new findings, of which some were discovered in paralell elsewhere (no biggie, confirmation is a good thing). However, since most of the book takes the form of a tour of the topic of CA, textbook-style, I don't know that one or two original findings in the first 2/3 of the book would be a bad track record.
On the other hand, I've not seen the "complexity is" sort of hypothesis (at least not from the CA angle... it certainly cropped up in chaos work in the 80s and 90s) that he goes into in the last third. It's a bit of a wild-eyed theory, and there's not much more that you can do than say, "yep, that's your theory all right." In 50-100 years, I suspect we'll have better tools to use to measure phenomenon on a very large scale that could confirm some of his theory, but for now I simply concede that it's not an unreasonable way of looking at the world.
I forget if he spends any time on HOW you would measure such phenomenon.
Please cite your references here. I don't recall anything of the sort (not even anything that could be paraphrased as such).
"Sure, the guy's brilliant. That doesn't make him right all the time."
Nor does the fact that he wrote a book against your better judgement make him wrong or, as many have said, "a kook".
"he won't mind if people try to knock holes in said theory"
Again, have you posed any questions? Have you written a paper discussing the merits and flaws of his approach and recieved any kind of feedback at all from him? I would like to know if you're just refering to his responses to people who call him a "kook" while heckling him in an auditorium or if you're actually refering to any sort of scholarly discorse.
I disagree entirely. If he looked up this information, adopted from it, and then made it his own, then he should cite his sources certainly. But if he's done this work on his own, I see no problem with publishing a book on the topic. If someone else also covered that ground, great! That way we can see where the two diverge and zero in on those areas as possible problems in the theory.
Or were we not actually interested in the science involved so much as the bragging rights?
I disagree that the book can be boiled down to your paraphrase. I strongly disagree, and would ask you to cite, chapter by chapter how the book fails to be more than that. I got quite a lot more out of the book than that.
Again, I'm not supporting his conclusions. He is speculating grandly, and I'm not going to say that he's right or wrong, we'll know that in a few hundred years, I'm sure.
What I'm saying is that those who dismiss this guy as a kook seem not to be pressing forward any substantive analysis, or if they do, it's of the form "x, y and z from the book have been discussed elsewhere"... well, ok I see that, but that certainly does not ivalidate his conclusions.
I do think he should have used a different title. Too many folks get a religious zelotry sort of reaction whenever you apply the word science to any kind of work. I would say that the first 2-thirds of the book are good, hard math, explored with a solid scientific approach.
The last third is just conclusion and extrapolation. Ignore it if you like. I find it quite interesting.
Here's my concern. I HAVE read the book. Not cover-to-cover, but a good deal of it, and I have to say that it's pie-in-the-sky stuff. Still, what I see people doing is jumping on this guy and saying that he's a kook, asshole, and various other derogitory words without a single shred of substantive argument against his points!
Yes: he is arguing that, at a very high level, current scientific approaches to large systems are flawed. I understand that that's off-putting to many, but you can't expect such a broad change in perspective to be done a) from the bottom-up nor b) without substantial leaning on the research of others. It's a theory, and a big one at that, so you accept that it's there and you don't base any substantial work or other theory on it until it is beaten on a bit. All I see in these responses is An Old Kind of Science that has been practiced by entrenched organizations like churches for thousands of years.
This kind of knee-jerk ostricization of bright people with ideas is just plain wrong. Maybe he's wrong about the idea, but you don't smack a guy down just for writing his ideas down, you correct where wrong if you want to be helpful or ignore if you don't. Being rude just isn't called for.
He claims no where in the book that I could find that others had not also covered much of this ground, but I don't think it's at all fair to say that that makes this book wholely unoriginal or at all ignorable. After all, he's relying on a body of mathematics that has been carved out over the last 3000 years! I also don't think that it's fair to say that ideas that he shares in common with others were not his. He's a bright enough guy that I think it's quite possible that in 10 years of cloistering himself off and working on this, he re-invented a few wheels. So what? That should neither diminish him nor those who covered the same ground before or in parallel with him (I don't think any less of Liebnitz, even though I'm not sure how to spell his name;-)
Let's all just calm down, take a deep breath and try not to be the Church of Established Science for a moment.
Someone else already followed up, but let me just give you some perspecitive. I've worked in several software companies that made proprietary software. If I had to stop and think about how much proprietary software I rely on, I'd probably stay at home, lock the doors and cut the power, gas, phone and cable.
The software industry is undisciplined, reckless and outright dangerous. Open source software tends to be no better at first cut, but where concerns that a proprietary software company might drag their feet about in order to focus on new customer features aren't so easily ignored when your user-base a) has the code and b) might also be your peer developers. What this means is that your average mature open source project is far better designed and debugged than your average proprietary software product.
Want examples? Try looking at ssh and openssh. Try looking at IE and Mozilla. Even better: gcc and any modern C or C++ compiler.
As these products mature, they are forced to improve in visible ways, but when all that is visible is a binary... it's hard to get the right focus.
Re:What is the *source* of the "RMS" controversy?
on
Stallman Goes to India
·
· Score: 1
It is important to note that Stallman's contributions to Computer Science and to software distribution are very different from his politics on the topics.
Emacs, GCC and the GPL are massive accomplishments, and I'm grateful for these. That doesn't mean I have to respect his rants about GNU/Linux vs. Linux; his repeated villification of anyone who tries to build a business around free software; or his us-vs-them approach to most corporations.
Is he eccentric? I've watched the man pick his nose with his chopsticks over dim sum and dance the halls of MIT for no decernable reason. Yeah, he's eccentric, but that's not my concern, I just don't feel compelled to be divisive and intolerant of mainstream software just because I respect free software and appreciate Stallman's contributions. I think that there's room enough in the world for both, and to claim otherwise is most of the reason for the free software / communism analogies that are so popular.
Also, the idea of a man who finds it difficult not to chew on his hair doesn't see, like the ideal ambasador of ideas to India, but perhaps I'm being pettty there....
The line I would draw (and this probably intersects the law at exactly zero points) is this: does the new work take market-share away from the original due to the original's name-recognition. If you are really parodying, there should be little or no intersection. If you are labeling a competing product with a recognizable play on the competition's name in order to attract business... well, screw you.
No, I won't get used to it. I will use whatever SPF-competing system allows for one-time (or at least short-duration) authorization of senders. It's simply not rocket science.
Exactly! People who advocate SPF seem to have this vision of the net as a static place where I send mail from a machine on my desktop and work or at home, and that's it. I may have to send mail from the control panel of an elevator that I have never even seen as me. Really. What I need is to have a standard way that that elevator control panel can warn my domain that "hey, I am spoofing mail as you, and that's ok because I have some magic secret". Once that protocol is in place, THEN something like the SPF transitional period begins to make sense (basically a period of time that you assume people wont have the tools installed, so you just issue warnings in the headers, while everyone starts installing the handshaking software). SPF can effectively NEVER leave transition because it has no mechanism to cover this case.
"Glib needs to stop duplicating functions and types that are already part of C. If you need strictly sized types C99 added a bunch of extra types for you to use; int16_t"
WOAH! You cannot rely on C9x support on every patform! Glib uses its own type naming conventions to avoid the morass of semi-standards and then identifies what you actually have at compile-time. This is the right way to go for a library that sets out to fundamentally re-architect the way you interface to the language (e.g. to put a real event, exception, threading and object model between you and the raw system), where it would NOT be the right choice for a lib that provided some specific niche of functionality.
All Gnome and Gnome-supporting software is designed to be portable. The problem is that "portable" != "ported". Compiler differences, subtle changes in libraries, etc can lead to platform differences that you need to account for. It might be good, for example, to rely on always having a SSL library around on modern systems, but on legacy OSes, you can't assume that at all! Same goes for assuming that C9x features are supported, and don't get me started on the difficulties in finding GCC dependencies without actually doing a port.
These things are all hurdles that you must overcome, and it's NEVER as easy as "just follow the standards".
There is no glue code required for a C++ program to use a C program, but there is (potentially extensive) glue required before I can use a C++ API from C. Let's just assume that we're not even talking about C++ that dips deeply into the STL, since then it gets even harder.
Not saying that you don't want to write C++ that is C++ rather than C-compiled-by-a-C++-compiler, I'm just saying that you introduce a layer of diffuculty in using libraries.
I use Perl a lot, so I understand when and where you want to decide that you don't care about all of that, and you just want to write the code fast.
"Too bad it's got portability issues"... "some people actually enjoy running Tru64/Irix/HP-UX"
And some people enjoy running VMS or Windows, but if it don't compile on your platform, I suggest you offer to help port it. NO software ports to every platform out-of-the-box, even (actually, perhaps especially) Java software.
On the other hand glib works fine on Solaris... in fact Sun ships glib WITH Solaris. Perhaps you should switch to Solaris, or ask your vendor why they don't support modern cross-platform libraries like glib that run on their competitor's boxes...
A C++ program can rely on a glib-based C library just as easily (perhaps somewhat easier, due to the consistent object model) as on any other C library. There is no problem here at all.
"Using Glib in KDE is pointless"
Using glib in DBUS is not however, and using DBUS in KDE is not... moot point. Also keep in mind that KDE's reliance on C++ and C++'s platform difficulties (SOME of which went away with the finalization of the ANSI standard a few years back) was exactly the reason that Sun had to choose Gnome as their desktop, even though they prefered KDE at the time. They had to support two compilers though, and if you can't lock customers into a compiler, C is the only way to fly (Java is as close as it gets otherwise, and it might be ok after another decade or two to mature).
I'm not language zeloting here... I see the value of C++ accedemically, but building software in it DOES cut you off from the rest of the world in the sense that the many, many thousands of C-based software projects and products in the world then have a hard time making any use of you at all.
The STL is not a C library, it's a C++ library, and on my box STL is over a megabyte and glib is under 200k, so I would suggest that you program in C and save yourself some overhead.
"Now, if you knew SPF, you would recognize that the last bit -- ?all"
Hate to sound snide, but if you knew SPF you would recognize that as a transitional setting, which the SPF specs suggest you set a hard cuttoff date around.
SPF's failing, as far as I can tell is that there is no dynamic authentication capability for a client out in space that wants to send mail "from" all of the 20 or so domains that that user had addresses with (e.g. my spamcop, personal, aol, work, oss project and other addresses). I don't want to go hunt down a server that will talk to me for mail origination for EVERY ONE of these domains... I just want a way to tell their servers, "hey, I just sent a message from your domain to joe@example.com, heads up" and have the right thing happen. There should then be a way for a server to say, "heya, I just got mail from your domain to my joe@example.com address... that yoy?" It needs to be message-by-message like this, and if that sounds like a lot of overhead... I GUARANTEE you that it is less than handling bounces for every virus message ever crafted in your name....
IMHO, every major C project should use glib. It is fairly lightweight and provides a lot of the features that C programmers end up carrying around anyway. It's certainly not a Gnome thing in the strictest sense. You could ship glib with KDE and have no dependence on Gnome, GTK+ or anything else like that.
Glib 2.0 also includes GObject, the core object system on which Gtk+ and Gnome are based, though again, you could write grep using these objects, they're not graphics-specific.
Just use Gnutella. gtk-gnutella's recent version includes most of the modern features of all of these commercial apps while a) being open source b) not including spyware c) running under Linux and other free-as-in-whathaveyou operating systems.
There are also other good clients for POSIX-like systems running X, but gtk-gnutella is the one I've been using and the last few versions are great. I pulled down the isos for Fedora Core 1.0 in a fraction of the time the ftp servers were choking it up to me....
Slashdot Mirror
Ego doesn't bother me. I've met a fairly large number of people with egos who entered the room long before they did, and I can say with some certainty that such people tend to be no more or less correct than you would find in the general population.
That said, thanks for the level-headed response.
Microsoft is not getting into it.
It is very important to remember that Microsoft (maker of Windows(tm) operating system) is a very different beast from Microsoft Research. MSR is a true research organization, and assuming that work done there will in any way translate into Microssoft products is a long-shot at best. Probably the most likely outcome of this kind of work is that some of the technology will be fed back into parts of various MS products.
I would think that if any part of his book speculated that such efficiency would be derived from CA, then that would be a reasonable request. That's not what the book is about. Give it a read.
I don't think anything you said is unreasonable.
I might disagree with some of it, but that's not the point. You've issued some very reasonable concerns about the BOOK and its CONTENT. That's fine.
What I was upset about were ad hominem attacks agaist Wolfram of the form "it wasn't peer reviewed, so he's a [kook,wacko,asshole,etc]" (yeah, someone actually called him an asshole because he was too egotistical and didn't accept peer review, though it was in another thread in this article).
I don't agree with his conclusions. I think that ultimately they will be shown to be a bit too simplistic, even if they're partially accurate. But, I refuse to write the man off as mentally abberant just because he a) wrote a book and b) didn't find every single reference to every other person who has covered the same ground before writing it. It's NOT an accedmic paper, and should not be treated as such. It's a very good book about a particular field of mathematics and an interesting treatment of one man's view of their impact on metaphysics.
"Do you know what "peer review" means?"
Why yes, I do.
"There's a reason why he is widely mocked."
Yes there is. The man wrote a book.
"You can't get taken seriously as an academic unless others get the chance to review your work"
That's just wrong, and only true of extremists.
What is and SHOULD be true is that if your work is not peer reviewed it is not accepted as mainstream science unless it is borne out by some external source later (Fermat's Last Theorem comes to mind). That is, you shouldn't go citing it as an authoritative source of anything, nor should people feel the need to defend new work on the basis of information found therein.
But to say that a PERSON cannot be taken seriously because they wrote a book is just silly. Was he a "kook" and/or "mocked" before he wrote a book? This guy isn't a university researcher out looking for handouts from the government. He's just a guy with some good ideas (and some far-fetched conclusions that may or may not have any bearing on reality). To call him a kook for publishing a book on these topics is the worst kind of elitism, and outright damaging to those who might follow in his footsteps and share what they have to say.
I'm a child of the information age (barely, I'm almost too old for it), and my feeling is that everyone should share the bounty of their thoughts. If you want to fit into some particular community, fine, you respect that community's norms, but if you want to share, go for it.
This guy, far from being a kook, is actually quite intellegent and rational. Yes, he makes some bold stabs in his conclusions, but that's not kooky, it's just intellectually risky. I don't see anything in that book that claims he's sure he's right or where he gives the written equivalent of the finger to anyone who disagrees.
The impression that I got was that there were a few relatively new findings, of which some were discovered in paralell elsewhere (no biggie, confirmation is a good thing). However, since most of the book takes the form of a tour of the topic of CA, textbook-style, I don't know that one or two original findings in the first 2/3 of the book would be a bad track record.
On the other hand, I've not seen the "complexity is" sort of hypothesis (at least not from the CA angle... it certainly cropped up in chaos work in the 80s and 90s) that he goes into in the last third. It's a bit of a wild-eyed theory, and there's not much more that you can do than say, "yep, that's your theory all right." In 50-100 years, I suspect we'll have better tools to use to measure phenomenon on a very large scale that could confirm some of his theory, but for now I simply concede that it's not an unreasonable way of looking at the world.
I forget if he spends any time on HOW you would measure such phenomenon.
"He refuses peer review"
That is to say, "he wrote a book".
"I'm right, everyone else is wrong"
Please cite your references here. I don't recall anything of the sort (not even anything that could be paraphrased as such).
"Sure, the guy's brilliant. That doesn't make him right all the time."
Nor does the fact that he wrote a book against your better judgement make him wrong or, as many have said, "a kook".
"he won't mind if people try to knock holes in said theory"
Again, have you posed any questions? Have you written a paper discussing the merits and flaws of his approach and recieved any kind of feedback at all from him? I would like to know if you're just refering to his responses to people who call him a "kook" while heckling him in an auditorium or if you're actually refering to any sort of scholarly discorse.
I disagree entirely. If he looked up this information, adopted from it, and then made it his own, then he should cite his sources certainly. But if he's done this work on his own, I see no problem with publishing a book on the topic. If someone else also covered that ground, great! That way we can see where the two diverge and zero in on those areas as possible problems in the theory.
Or were we not actually interested in the science involved so much as the bragging rights?
I disagree that the book can be boiled down to your paraphrase. I strongly disagree, and would ask you to cite, chapter by chapter how the book fails to be more than that. I got quite a lot more out of the book than that.
Again, I'm not supporting his conclusions. He is speculating grandly, and I'm not going to say that he's right or wrong, we'll know that in a few hundred years, I'm sure.
What I'm saying is that those who dismiss this guy as a kook seem not to be pressing forward any substantive analysis, or if they do, it's of the form "x, y and z from the book have been discussed elsewhere"... well, ok I see that, but that certainly does not ivalidate his conclusions.
I do think he should have used a different title. Too many folks get a religious zelotry sort of reaction whenever you apply the word science to any kind of work. I would say that the first 2-thirds of the book are good, hard math, explored with a solid scientific approach.
The last third is just conclusion and extrapolation. Ignore it if you like. I find it quite interesting.
Here's my concern. I HAVE read the book. Not cover-to-cover, but a good deal of it, and I have to say that it's pie-in-the-sky stuff. Still, what I see people doing is jumping on this guy and saying that he's a kook, asshole, and various other derogitory words without a single shred of substantive argument against his points!
;-)
Yes: he is arguing that, at a very high level, current scientific approaches to large systems are flawed. I understand that that's off-putting to many, but you can't expect such a broad change in perspective to be done a) from the bottom-up nor b) without substantial leaning on the research of others. It's a theory, and a big one at that, so you accept that it's there and you don't base any substantial work or other theory on it until it is beaten on a bit. All I see in these responses is An Old Kind of Science that has been practiced by entrenched organizations like churches for thousands of years.
This kind of knee-jerk ostricization of bright people with ideas is just plain wrong. Maybe he's wrong about the idea, but you don't smack a guy down just for writing his ideas down, you correct where wrong if you want to be helpful or ignore if you don't. Being rude just isn't called for.
He claims no where in the book that I could find that others had not also covered much of this ground, but I don't think it's at all fair to say that that makes this book wholely unoriginal or at all ignorable. After all, he's relying on a body of mathematics that has been carved out over the last 3000 years! I also don't think that it's fair to say that ideas that he shares in common with others were not his. He's a bright enough guy that I think it's quite possible that in 10 years of cloistering himself off and working on this, he re-invented a few wheels. So what? That should neither diminish him nor those who covered the same ground before or in parallel with him (I don't think any less of Liebnitz, even though I'm not sure how to spell his name
Let's all just calm down, take a deep breath and try not to be the Church of Established Science for a moment.
Someone else already followed up, but let me just give you some perspecitive. I've worked in several software companies that made proprietary software. If I had to stop and think about how much proprietary software I rely on, I'd probably stay at home, lock the doors and cut the power, gas, phone and cable.
The software industry is undisciplined, reckless and outright dangerous. Open source software tends to be no better at first cut, but where concerns that a proprietary software company might drag their feet about in order to focus on new customer features aren't so easily ignored when your user-base a) has the code and b) might also be your peer developers. What this means is that your average mature open source project is far better designed and debugged than your average proprietary software product.
Want examples? Try looking at ssh and openssh. Try looking at IE and Mozilla. Even better: gcc and any modern C or C++ compiler.
As these products mature, they are forced to improve in visible ways, but when all that is visible is a binary... it's hard to get the right focus.
It is important to note that Stallman's contributions to Computer Science and to software distribution are very different from his politics on the topics.
Emacs, GCC and the GPL are massive accomplishments, and I'm grateful for these. That doesn't mean I have to respect his rants about GNU/Linux vs. Linux; his repeated villification of anyone who tries to build a business around free software; or his us-vs-them approach to most corporations.
Is he eccentric? I've watched the man pick his nose with his chopsticks over dim sum and dance the halls of MIT for no decernable reason. Yeah, he's eccentric, but that's not my concern, I just don't feel compelled to be divisive and intolerant of mainstream software just because I respect free software and appreciate Stallman's contributions. I think that there's room enough in the world for both, and to claim otherwise is most of the reason for the free software / communism analogies that are so popular.
Also, the idea of a man who finds it difficult not to chew on his hair doesn't see, like the ideal ambasador of ideas to India, but perhaps I'm being pettty there....
The line I would draw (and this probably intersects the law at exactly zero points) is this: does the new work take market-share away from the original due to the original's name-recognition. If you are really parodying, there should be little or no intersection. If you are labeling a competing product with a recognizable play on the competition's name in order to attract business... well, screw you.
No, I won't get used to it. I will use whatever SPF-competing system allows for one-time (or at least short-duration) authorization of senders. It's simply not rocket science.
Exactly! People who advocate SPF seem to have this vision of the net as a static place where I send mail from a machine on my desktop and work or at home, and that's it. I may have to send mail from the control panel of an elevator that I have never even seen as me. Really. What I need is to have a standard way that that elevator control panel can warn my domain that "hey, I am spoofing mail as you, and that's ok because I have some magic secret". Once that protocol is in place, THEN something like the SPF transitional period
begins to make sense (basically a period of time that you assume people wont have the tools installed, so you just issue warnings in the headers, while everyone starts installing the handshaking software). SPF can effectively NEVER leave transition because it has no mechanism to cover this case.
"Glib needs to stop duplicating functions and types that are already part of C. If you need strictly sized types C99 added a bunch of extra types for you to use; int16_t"
WOAH! You cannot rely on C9x support on every patform! Glib uses its own type naming conventions to avoid the morass of semi-standards and then identifies what you actually have at compile-time. This is the right way to go for a library that sets out to fundamentally re-architect the way you
interface to the language (e.g. to put a real event, exception, threading and object model between you and the raw system), where it would NOT be the right choice for a lib that provided some specific niche of functionality.
All Gnome and Gnome-supporting software is designed to be portable. The problem is that "portable" != "ported". Compiler differences, subtle changes in libraries, etc can lead to platform differences that you need to account for. It might be good, for example, to rely on always having a SSL library around on modern systems, but on legacy OSes, you can't assume that at all! Same goes for assuming that C9x features are supported, and don't get me started on the difficulties in finding GCC dependencies without actually doing a port.
These things are all hurdles that you must overcome, and it's NEVER as easy as "just follow the standards".
There is no glue code required for a C++ program to use a C program, but there is (potentially extensive) glue required before I can use a C++ API from C. Let's just assume that we're not even talking about C++ that dips deeply into the STL, since then it gets even harder.
Not saying that you don't want to write C++ that is C++ rather than C-compiled-by-a-C++-compiler, I'm just saying that you introduce a layer of diffuculty in using libraries.
I use Perl a lot, so I understand when and where you want to decide that you don't care about all of that, and you just want to write the code fast.
"Too bad it's got portability issues" ... "some people actually enjoy running Tru64/Irix/HP-UX"
And some people enjoy running VMS or Windows, but if it don't compile on your platform, I suggest you offer to help port it. NO software ports to every platform out-of-the-box, even (actually, perhaps especially) Java software.
On the other hand glib works fine on Solaris... in fact Sun ships glib WITH Solaris. Perhaps you should switch to Solaris, or ask your vendor why they don't support modern cross-platform libraries like glib that run on their competitor's boxes...
A C++ program can rely on a glib-based C library just as easily (perhaps somewhat easier, due to the consistent object model) as on any other C library. There is no problem here at all.
"Using Glib in KDE is pointless"
Using glib in DBUS is not however, and using DBUS in KDE is not... moot point. Also keep in mind that KDE's reliance on C++ and C++'s platform difficulties (SOME of which went away with the finalization of the ANSI standard a few years back) was exactly the reason that Sun had to choose Gnome as their desktop, even though they prefered KDE at the time. They had to support two compilers though, and if you can't lock customers into a compiler, C is the only way to fly (Java is as close as it gets otherwise, and it might be ok after another decade or two to mature).
I'm not language zeloting here... I see the value of C++ accedemically, but building software in it DOES cut you off from the rest of the world in the sense that the many, many thousands of C-based software projects and products in the world then have a hard time making any use of you at all.
The STL is not a C library, it's a C++ library, and on my box STL is over a megabyte and glib is under 200k, so I would suggest that you program in C and save yourself some overhead.
"Now, if you knew SPF, you would recognize that the last bit -- ?all"
Hate to sound snide, but if you knew SPF you would recognize that as a transitional setting, which the SPF specs suggest you set a hard cuttoff date around.
SPF's failing, as far as I can tell is that there is no dynamic authentication capability for a client out in space that wants to send mail "from" all of the 20 or so domains that that user had addresses with (e.g. my spamcop, personal, aol, work, oss project and other addresses). I don't want to go hunt down a server that will talk to me for mail origination for EVERY ONE of these domains... I just want a way to tell their servers, "hey, I just sent a message from your domain to joe@example.com, heads up" and have the right thing happen. There should then be a way for a server to say, "heya, I just got mail from your domain to my joe@example.com address... that yoy?" It needs to be message-by-message like this, and if that sounds like a lot of overhead... I GUARANTEE you that it is less than handling bounces for every virus message ever crafted in your name....
IMHO, every major C project should use glib. It is fairly lightweight and provides a lot of the features that C programmers end up carrying around anyway. It's certainly not a Gnome thing in the strictest sense. You could ship glib with KDE and have no dependence on Gnome, GTK+ or anything else like that.
Glib 2.0 also includes GObject, the core object system on which Gtk+ and Gnome are based, though again, you could write grep using these objects, they're not graphics-specific.
And long-term I think it's best that P2P networks be based on open standards. Fast Track is anything but, from what I can tell.
Just use Gnutella. gtk-gnutella's recent version includes most of the modern features of all of these commercial apps while a) being open source b) not including spyware c) running under Linux and other free-as-in-whathaveyou operating systems.
There are also other good clients for POSIX-like systems running X, but gtk-gnutella is the one I've been using and the last few versions are great. I pulled down the isos for Fedora Core 1.0 in a fraction of the time the ftp servers were choking it up to me....