I'll make it easy for you. It's the two minute litmus test. Even easier -- I'll give you the pseudo-C code: Task: compute n! for n >= 1000.
In Scheme (Gambit 4.2.8, using infix):
int factorial(int n) {
if (n (load "f") "/home/user/f.o1" >(factorial 1000) 4023...0000
Your challenge? Write a C version in two minutes, tested and compiled. Now, as the final icing, run the C version on smaller numbers, and compare the performance -- did you forget to compile in small integer versions? (try factorial(12) a million times).
I'll wait (another two minutes). Compare the performance against the LISP version. Did you have to write two versions -- one for big integers and one for small integers? That is pretty well the only way to keep a speed advantage... I hope you wrote it that way. Did you remember to put in 32/64 bit conditionals to retain your advantage on a 64 bit platform?
I think your C code now looks like this (it should):
#define FACT_LIMIT 12 -- for 32 bit int type, I don't know what the cutoff is for 64 bit. #include bignum.h -- I don't want to bother with quoting assume angle brackets/* This only gets executed a maximum of FACT_LIMIT times; leave it recursive */ int fact_integer(int n) {
if (n = 0) {
return 1;
} else {
return n * factorial(n - 1);
} }/* May wish to rewrite to an iterative form */ bignum factorial(bignum n) {
if (compare_lt(n, FACT_LIMIT)) {
return int_to_bignum(fact_integer(bignum_to_int(n)));
}
return bignum_mult(n, bignum_dec(n)); }
You choose the bignum package to use. Or, for more fun, write it yourself. If you wrote it yourself, you remembered to switch to FFT style multiplication at bigger sizes? Or Karatsuba?
Now, we have only coded to a recursive form, but, since bigints are not first-class in C, we don't know about memory reclamation (leakage). I hope you know the gmp library, or can roll up a gee-whiz allocator on your own. The gmp library would be cheating, by the way -- YOU DID CLAIM YOUR IMPLEMENTATION IN C.
If recursion is viewed as a problem, the Gambit-C version can be recoded as:
int factorial(int n) {
int i;
int a;
if (n = 0) {
1;
} else {
a = 1;
for (i = 1; i = n; ++i) {
a *= i;
}
a;
} }
I am sure that something equivalent can be done in the C version. But the normal flow of control stuff doesn't know about bignums. We COULD make the incoming parameter an int, I guess... which works for factorial() but may not be as workable for other functions.
Answers: - gmp does better than Gambit-C on bigint multiply, using FFTs. - breaking the result into two separate functions is needed for C to come ahead. - yes, C is faster, at the expense of a lot more programming. - if I want to, I can simply drop C code into my Gambit-C program on an as-needed basis. The Gambit-C code still looks a
whole lot cleaner than the C version, and ties it for small integer performance. The bigint performance is still a "win" for
gmp, but I can use THAT package directly as well in Gambit-C.
Win: - Gambit-C. The prototype was finished to spec in two minutes. Optimizations can wait and are as implementable as in a native
C version.
Note that there is nothing non-LISP about Gambit. I chose it for this example because it has a infix notation available and is comfortable for C developers. There is nothing particularly special about the example either. It was chosen because it was small.
" Programming a 1 GHz Pentium [fastest, 1000$, 2002 PC] in Perl is like programming a 10 MHz something [overclocked, 50$, 1982 Z80-based ZX Spectrum] in FORTRAN!
This is no big surprise as we are looking at the classical trade-off between the power of a language (allowing the programmer to express something in a compact way) and the performace of an implementation (which classically was related to the language being close to the machine representation).
However, the huge exception is CommonLisp. Lisp is the most powerful language that is, representing the classical extreme choice for expressive power instead of efficient implementation.... For new projects in which one is free to choose the language, one should choose Lisp."
Of course C is an extremely primitive language. It is possible to (on current hardware) write a C program to best most other systems. But, a proper program is much more difficult!
Microsoft LOSES market share if you install something else over top?
I would beg to differ -- all machines are still counted as Windows(tm) machines. This allows demonstration of the ABSOLUTE and CRUSHING numerical superiority of Windows.
Must be, the number of attempts per second makes no sense. I get a COMPLETELY different answer based on similar "data" from his article:
Even if the password is "trivially" generated:
symbol word symbol word symbol: eg. _orange*bear#
would require 10,000 attempts per second to crack under these circumstances (roughly). Assuming that the defender uses ONLY this exact pattern:
The symbol characters add 4 bits each, or 1000 multiplier for 3 (very rough). Choose two words from the dictionary, say limited to 10,000 choices each (the author states that it could be chosen from a set of 50,000). 100,000,000,000 possibilities to run through, If 100 days given, 1,000,000,000 have to examined per day, 86,400 seconds per day, or just over 10,000/second. Contrast to 65/minute.
And that assumes a VERY limited password.
I should download and review this, but I really don't care enough.
But I will try to guess where that number could come from:
Two words together, nothing else. 100,000,000 combinations, 1,000,000 per day, or just over 10 per second.
A single dictionary word, followed by a number. 100,000 combinations. Finally, our attack rate can be 1 per second.
"In the bedroom". No. But what I need is POSIX compatible systems. I am happy enough with Linux, although I prefer Solaris.
As long as the operating environment supports what I need, I am willing to entertain it. Windows(tm)? Last time I looked, Microsoft was busy deprecating getcwd(). Apparently, ISO C++ insists on _getcwd(). I didn't know that Microsoft was particularly ISO compliant either (or maybe I'm stupid). MSVC (my version - 7), doesn't even support stdint.h. I guess I COULD rewrite a crapload of code, but it isn't happening.
But then, I guess my needs aren't "typical". I like OpenOffice.org presenter, with the current slide/next slide/clock/notes view. If I present and have to use the Microsoft product, I need to remember to find a suitable clock. But then, I guess I am not typical. I need a platform that integrates into Unix platforms, and supports NIS and NFS. But then, I guess I am not typical.
I have tried Windows and other Microsoft products. They STILL aren't suitable for me. But I am ever hopeful...
And what, were you using then? Windows 3.11, I presume. In which case, Linux 1.0 (released March, 1994) offered a 32 bit system with memory protection and preemption.
So, tell me, did you ENJOY running EMM386.EXE?
I guess you could have been running Windows NT 3.1 (or 3.5, released September 1994). That also offered a 32 bit system with memory protection and preemption. It also cost $280 back then. Linux? Free.
OpenOffice.org is "amazing". I have no freaking clue what Microsoft Office(tm) does, but I do know some parts of it having been forced into using those parts.
The parts that are important to me are:
Giving presentations: OpenOffice.org has an amazing control panel that shows current/next slide, time, and speakers notes on a separate monitor. What a godsend for presenters!
Editing PDFs: OpenOffice.org can do it.
Saving PDFs: OpenOffice.org can do it. I believe that MS Word(tm) can finally do this.
Can see fonts in menu: I believe later versions of MS Word(tm) copied this feature.
So the point of making it superior isn't actually true (Linux WAS ALSO clearly superior to Windows 3.1, 3.11 WfW, Windows 95 and Windows 98). "They" may not come, anyway.
I used to think that I should evangelize (when I was younger, and not as wise) -- now I just say "You will know if you want to switch, and why. Do whatever makes you happy."
Assume an exploit on a platform, and that the exploit can spread. It will spread on systems of the same platform, and it doesn't matter WHAT the percentages are. If an exploit hits (say) 70% of Windows(tm) boxes, it will hit 70%. Whether the Windows boxes are 90% of the population, or 30%.
But, if Windows has 90%, it has numerically greater numbers, which is of desire (for ddos attacks, etc.). Supposedly, anyway (we'll be back to that in a moment) so greater number of blackhats look for exploits on that platform. They don't because Mac, Linux, Solaris, AIX, HP/UX, BSD etc. have such low figures that it wouldn't be worth it!
Ok, this makes sense for credit card data, and the pride of pwning a large number of systems. Doesn't make as much sense for bandwidth. Most of the boxes that have access to the highest bandwidth run Unix (Solaris, AIX) or Linux.
So either the Unix/Linux boxes were not considered valuable targets, or it was beyond the blackhats expertise. XP was shooting fish in a barrel; Vista is more promising on a security level. Note that social engineering "trojans" don't work nearly as effectively on those high-bandwidth systems (typically headless, with professional admins).
In answer to your question, in the 30/30/30 world, the Windows (at XP level) would still have the largest number of infections. Mac would be far less (simply no open ports), but there is a possibility of a Mac trojan (and TFA relates to that exactly). Linux? Also has an interesting infection vector for trojans based on desktop files. Neither Mac nor Linux would allow an easy automatic virus, although Linux would be/slightly/ easier. Once in, the Mac box is easier to exploit further.
With Vista, the automatic virus propagation is mitigated, because, even if/when the box is "exploited" via a service, the exploit code cannot reliably do anything (at least this month).
Sure, TeX will allow you to "program" documents, but I doubt that it what you meant. TeX sources files "written" as standard ASCII files. Which, in a Unix environment, means that the standard environment applies. You can use awk/sed/perl etc. to "automate" your documents.
Word(tm) and OpenOffice.org (because it wants to be "competitive" with Word) both incorporate their OWN programming language. About the only good that I have seen come of that idea is that Word could be used as an entry-level programming environment for Windows(tm). That this idea could be (seriously) floated boggles my mind.
Given that the ENTIRE purpose of the Unix environment is exactly that -- file and document automation, and drills down to that end with laser focus, duplicating that functionality in the "word processor" is stupid. The purpose of TeX is to apply that same laser focus to producing those "immaculate looking documents".
Separating functionality allows the parts themselves to be clearer and, as a corollary, more defect free.
As to standards -- the POSIX environment is the standard. TeX uses ASCII which is the standard. The TeX processor itself is arguably stable enough over 30 years to be considered standard. Which Word standard are you referring to? The one ratified last year I presume.
What is a "Word form system"? I am presuming you are either making a form in Word, which wouldn't be a system, or (more likely) using Word to fill out forms and file the results. In other words, using a Word Processor as a data entry system. Isn't that a bit like mowing a lawn with a shaving razor? To Microsoft's credit, once someone gets a thing like that working, it will be VERY painful to port to anything else (moving is simply impossible).
"We could easily construct a wishlist or consumer apps for Linux. My list would include Adobe's Photoshop Elements and Premiere Elements, as well as their professional graphics products (Dreamweaver, Illustrator, InDesign, Photoshop). I'd also add Intuit's Quicken and TurboTax, Roxio Creator or Toast, a Linux equivalent of WinDVD or CyberDVD, drawing programs like Visio and CorelDraw, a painting program, a website creation program like RapidWeaver or Freeway Express, and much more. There's also a very long list of educational titles and game titles that are almost entirely absent from Linux."
I don't know what Photoshop Elements is. I use gnucash for financial management. gnomebaker for making CDs. WinDVD? I don't know -- my linux laptop came with something that plays DVDs. I use dia for drawing, and kolourpaint for painting. I haven't bothered with a website creation program. As for educational and games - generally, I don't bother. The kids use flash games on the web, as well as educational resources.
All of this software came with my current laptop "integrated" (pre-installed, or manufacturer supported download). It's an Acer Aspire One, running the horribly named "Linpus" system. Adobe Reader and Flash pre-installed as well.
I did install an mplayer codec pack to allow (media-du-jour) to play.
Of course, having the list is of use to me as well; I need to know the Windows equivalents occasionally.
You are complaining about Printing, Networking, Laptop features.
What were you using in 1992? Windows 3.11, WfW? Or, did you use a Mac? Maybe OS/2? Presuming you were using "standard" architectures, because that was all Linux ran on back then...
Back then, Linux was far and away the technical lead. Except, of course, when compared to OS/2, or SCO. (My "rant" apologies if you were on one of those platforms).
Linux offered REAL processes and preemption, and REAL isolation. It was able to support real servers, and REAL applications. Windows? Mac? Not so much. Simply holding down the mouse button could crash networks with MacOS.
Linux proceeded to develop, as did Windows(tm) and Mac(tm). Apple eventually introduced OS X, and Microsoft the NT based kernels. Bringing their stuff to parity.
As a Unix (Linux) user, I didn't really bother with USING Windows(tm). It didn't offer anything and wasn't compatible with the applications I wanted to run. Finally, Windows(tm) caught up, as did Apple with OS X. Still, there are missing pieces (NIS, NFS others). At least Apple owns CUPS now.
As to printing: (to give a specific)
The Unix model for the past twenty years has been to have the applications generate Postscript. That will go to a standard LPD, and may be sent to a different machine. If the printer being used is "inferior", GS may be used to generate a raster for it. This is, of course, isolated from the application.
CUPS adds printer publishing (don't need to define the print queues anymore).
And that's all there is to it.
Windows(tm) breaks this model. GDI is used to rasterize and drive the printer. Completely different, and in my opinion, wrong and broken. CUPS does try to accommodate the Windows(tm) model, but it is, at best, a poor emulation. Of a broken system, so it doesn't matter to me anyway.
You can try to argue that a GDI based model is superior to a Postscript model -- but Postscript has far better resolution and device independence. It IS the standard for Unix printing. Which is NOT Windows(tm) printing.
Yes, I am one of those crusty "Here's a Nickel, Kid, Get yourself a better computer" guys.
Yes, Windows(tm) and the Mac OS have (for the most part) finally caught up. But they don't really offer much extra; except for some specific applications.
If you NEED one of those applications you would be better served with the system that actually supports it. Personally? Since I am an old-hat Unix/Linux user, I don't have any such applications.
Um... no. If I buy a copy, I cannot return it. And I will not "pirate" it. So, it's not happening. If it doesn't run on Solaris and Linux, I can't use it. So, it's not happening.
Which means I use OpenOffice.org. With menus. If I have to use MS Office(tm) I, of course, go nuts. Never had the "training" or familiarity.
So, I say, "stuff it". If it isn't OpenOffice.org compatible, take a hike.
I am currently engaged with a client. For a number of reasons, we had to accept "unlimited liability" in the case of information leakage (and that could easily run into the millions).
If I have a VPN connection linked up to the internal network, I cannot tolerate other external connections. Whether or not "malware" is transmitted internally is not my concern (and it probably isn't an issue; I use neither Windows(tm) or OS X).
Further, I don't store "sensitive" client data on my machine; other data is stored encrypted. Even if I am physically at the client site, there is a separation with network cables -- some have external access, and some are internal. To bypass this, I guess I could connect TWO cables, but my machine only has a single network port preventing this from occurring accidentally.
This VPN behavior is useful to those of us who need an "air-gap" or something arguably similar. Yes, I use VPNs like this, and I guess you could call me paranoid.
It would SEPARATE hardware innovation from software innovation. Let the hardware companies do their own thing, and supply specifications.
What may happen is that time to market increases. Basically, the new hardware has to be seeded (or purchased) by software developers and support organizations before general adoption.
Large software houses would be more conservative with support, while smaller, less risk averse houses would offer earlier support.
In other words, Microsoft Windows(tm) would no longer be the FIRST platform to be supported. As a corollary, it would not be the ONLY platform supported.
But, there will be vendors developing drivers into the Microsoft ecosystem as well.
Cost of hardware should also be reduced. Users of BSD, Linux, and other "free" platforms won't need to pay for the price of Windows(tm) drivers.
Commercial driver support will be possible, and not just as contracting to hardware companies (which is the current situation). Since hardware specification would be available, low level software innovation will again be possible.
It would not stifle innovation. Indeed, the opposite would happen.
Canon sells cameras, not software. The software is added as an "extra". Since the software IS supplied, Canon tries to support it, and limits the supported platforms (trying to control costs).
If Canon were liable for the "licensed, not sold, software", they would have to do something about the software. Either improve its quality (and cost), or, more simply, not supply software. The camera would then be hardware conforming to a specification. Look at modems, printers, and other peripherals in the pre-Windows 9x era. The devices came with clear technical interface manuals, or compliance to standards. Others were expected to write the "drivers" to make the devices useful. Sometimes the vendor would supply software, but more as an example, and almost always with source code (the example should be useful).
[Sidenote: This practice was even extended into the early PC computers. Apple published source for the Apple ][ ROM, KIM-1, SYM-1 had the source supplied, and even the IBM PC had the BIOS listing]
This would again allow platform variation. The smaller platform suppliers would be able to support new peripherals FASTER than the larger vendors, thus, in my opinion, (re)balancing the software landscape.
My internet is already balkanized. As much as possible. Because I don't trust you. It's MY network, and I'll allow YOUR network access if and when I trust your network.
So far, my trust network is only to two other networks. I know the admins personally. Quid Pro Quo, one of the admins is my outbound mail relay.
Yes, I'm probably paranoid. But really, everyone else gets minimal access and services.
I think you will find that quite a few people were already on the "cutting edge". Using OS/2 and such. Windows wasn't even an "also-ran", it was, at that time, a complete non-starter (1991-1996).
Linux is not the next thing after Windows -- by any measure it is a contemporary. The question that many have is "why doesn't Microsoft make Windows more compatible"?
It helps to revise history (from the winners perspective), of course. But, back in the early '90s, Linux wasn't in competition with Windows (Windows 3.1, WfW? Give me a break). It was in competition with SCO (and, to a lesser extent, BSD).
Windows was NOT even remotely the target -- it had already been surpassed (technology-wise). You could build a (reasonably reliable) server based on Linux 0.99. Try that with Windows 3.1 or Mac OS of the day!
Linux allowed easier experimentation with OSs than Minix, and BSD was caught up in some disputes. SCO was too expensive.
Wrong. The right of property gives the right to reproduce. After all, you can do anything you want with your property.
This right is limited by Copyright. ("takes it away")
Copyright is NOT the start of this reasoning -- it falls under.
The creator certainly has the right to not share. If the creator desired absolute control, she may decide not to share, or to apply Trade Secret or other (contract) provisions. But, if the work is shared, it becomes "open" in the sense that the creator is not deprived of her ownership by the fact of other ownership. (which distinguishes "Copyright violation" from "theft").
In Unix (Linux), there is a rigid placement for code --/bin and/sbin contain the OS utilities,/lib for neeed libraries, and/boot contains the OS itself (/kernel for Solaris, but the idea is the same)./usr/bin,/usr/sbin,/usr/lib et al contain OS utilities NOT NEED FOR BOOTING (/usr can be network mounted)./etc for OS configuration./usr/local and-or/opt for site specific programs. (Applications not supported by the OS vendor directly, or, in the case of Solaris, by a different group)
Easy to upgrade the OS. Replace the OS parts, and leave/usr/local,/opt and the/home (/export/home) directories alone. No particular complications. Everyone administering a box knows the score, and we don't have any problems. Of course,/etc changes are then the worst, but apps can be encouraged to go with/usr/local/etc for global configuration (this is rarely done, it turns out to not be a practical necessity).
As to what is installed? Different OSs use different conventions (obviously, and Microsoft uses their own). yum list in Fedora (for example) will give the list of packages, libraries and utilities currently installed via the vendor installer/updater (2373 installed packages on my netbook I am using to type this). It does behoove Microsoft to keep track of all of their own packages, beta to beta and to release (it would be part of bug tracking).
Given the experience in this area gained from 40 years of use, I would imagine the Windows Update procedure to be as finely honed, especially given the new security concerns of Microsoft in relation to Windows.
So it doesn't make sense to me that a beta to beta update isn't supported. This must be strictly a "scope limiting" move; but I don't think it reflects well on Microsoft and its position as an OS vendor.
"There is no inalienable right to use copyrighted material"
There sure is. The right of property. I assume that you are talking about US law, given the article to which you are commenting.
The Bill of Rights prohibits the federal government from depriving any person of life, liberty, or property, without due process of law.
A CD, book or DVD is simply property and you can use it however you want.
Of course, you are talking about Copyright, which, in this case refers to the right to make a copy of the property in question. Since the Bill of Rights trumps Copyright, Copyright cannot restrict what you do with it, outside of the (hopefully narrow) proscriptions specified and justified in the Copyright Act.
So, there is an inalienable right. (Well, technically, I guess you COULD argue that the Bill of Rights is not inalienable, but it should be so considered in the US).
What we are talking about is then the scope (or restrictions) placed on Copyright, and the reactions of Companies to overly broad interpretations of Copyright as promoted by cartels and not creators.
"Double the pipeline length...directly led to lower instructions per cycle"
Wouldn't twice as many instructions be worked on per cycle? Shouldn't this result in a an increase in instructions per cycle, for "properly" written code?
"Things usually go like this" -- Maybe, but I have used computers for a LONG time. It went more like this:
(Note that ALL details have happened to me, as listed, but not necessarily in the order presented. Because of my Windows XP "experience" I refuse to buy Windows Vista.).
Windows Fanboy: Windows XP is great, it does anything you need it to do!
Greybeard User: That's nice. The last time I tried it was Windows 98SE, and that version couldn't really walk and chew gum at the same time without dying. Maybe I'll try again.
Greybeard User: I tried to install in (after spending $200 on it) in order to run a TV tuner recorder, but it doesn't recognize the CD it just installed from.
WF: It needs drivers.
GU: Right, but it doesn't recognize the Network, or USB, and the drivers are too big to fit on a floppy...
WF: Just "slipstream" them.
GU: What? I don't HAVE another computer running Windows XP.
WF: You must be a Unix Loser!
GU: I think I'll try this "Fedora Core" thing. Hey! it works.
GU: [after many months pass]. Ok, I have this "Windows XP" running on a VMware instance. Why don't I have a NIS option for login? Don't ALL Operating Systems support that?
WF: You must be a Unix Loser. I don't know what that is!
GU: The command line sucks. Why can't I have a POSIX shell like Bash?
WF: It does not suck, you tool! And, you can use Cygwin to run Bash.
GU: And, it doesn't run the programs I have been using. Nothing works!
WF: Use Cygwin, Duh!
GU: What is this Cygwin you keep talking about?
WF: It lets you run Unix programs on Windows. RTFM noob.
GU: What manual? I bought Windows XP, so it didn't come with a paper manual and if you don't know what you are looking for, the on-line help isn't very helpful.
WF: Umm...
GU: If Windows is so great and does everything I need it to, why is something like Cygwin even necessary? And, even WITH Cygwin, I still can't use NIS and standard logins. And, why is Cygwin fork function so damn slow? My programs run like a tortoise in molasses.
WF: Because of tools like you! You are just a Unix Loser! Windows is great so just shut your hole!
Slashdot Mirror
Trolling sure sounds easy, but...
Gambit-C Scheme vs. C
I'll make it easy for you. It's the two minute litmus test. Even easier -- I'll give you the pseudo-C code:
Task: compute n! for n >= 1000.
In Scheme (Gambit 4.2.8, using infix):
int factorial(int n) {
if (n (load "f")
"/home/user/f.o1"
>(factorial 1000)
4023...0000
Your challenge? Write a C version in two minutes, tested and compiled. Now, as the final icing, run the C version on smaller numbers, and compare the performance -- did you forget to compile in small integer versions? (try factorial(12) a million times).
I'll wait (another two minutes). Compare the performance against the LISP version. Did you have to write two versions -- one for big integers and one for small integers? That is pretty well the only way to keep a speed advantage... I hope you wrote it that way. Did you remember to put in 32/64 bit conditionals to retain your advantage on a 64 bit platform?
I think your C code now looks like this (it should):
#define FACT_LIMIT 12 -- for 32 bit int type, I don't know what the cutoff is for 64 bit. /* This only gets executed a maximum of FACT_LIMIT times; leave it recursive */ /* May wish to rewrite to an iterative form */
#include bignum.h -- I don't want to bother with quoting assume angle brackets
int fact_integer(int n) {
if (n = 0) {
return 1;
} else {
return n * factorial(n - 1);
}
}
bignum factorial(bignum n) {
if (compare_lt(n, FACT_LIMIT)) {
return int_to_bignum(fact_integer(bignum_to_int(n)));
}
return bignum_mult(n, bignum_dec(n));
}
You choose the bignum package to use. Or, for more fun, write it yourself. If you wrote it yourself, you remembered to switch to FFT style multiplication at bigger sizes? Or Karatsuba?
Now, we have only coded to a recursive form, but, since bigints are not first-class in C, we don't know about memory reclamation (leakage). I hope you know the gmp library, or can roll up a gee-whiz allocator on your own. The gmp library would be cheating, by the way -- YOU DID CLAIM YOUR IMPLEMENTATION IN C.
If recursion is viewed as a problem, the Gambit-C version can be recoded as:
int factorial(int n) {
int i;
int a;
if (n = 0) {
1;
} else {
a = 1;
for (i = 1; i = n; ++i) {
a *= i;
}
a;
}
}
I am sure that something equivalent can be done in the C version. But the normal flow of control stuff doesn't know about bignums. We COULD make the incoming parameter an int, I guess... which works for factorial() but may not be as workable for other functions.
Answers:
- gmp does better than Gambit-C on bigint multiply, using FFTs.
- breaking the result into two separate functions is needed for C to come ahead.
- yes, C is faster, at the expense of a lot more programming.
- if I want to, I can simply drop C code into my Gambit-C program on an as-needed basis. The Gambit-C code still looks a
whole lot cleaner than the C version, and ties it for small integer performance. The bigint performance is still a "win" for
gmp, but I can use THAT package directly as well in Gambit-C.
Win:
- Gambit-C. The prototype was finished to spec in two minutes. Optimizations can wait and are as implementable as in a native
C version.
Note that there is nothing non-LISP about Gambit. I chose it for this example because it has a infix notation available and is comfortable for C developers. There is nothing particularly special about the example either. It was chosen because it was small.
Why would LISP be interpreted? (that is your implication, is it not?).
http://www.ffconsultancy.com/languages/ray_tracer/index.html
Scheme (unadorned) is 30% slower that g++.
http://dan.corlan.net/bench.html
Choice quotes from the last page:
" Programming a 1 GHz Pentium [fastest, 1000$, 2002 PC] in Perl is like programming a 10 MHz something [overclocked, 50$, 1982 Z80-based ZX Spectrum] in FORTRAN!
This is no big surprise as we are looking at the classical trade-off between the power of a language (allowing the programmer to express something in a compact way) and the performace of an implementation (which classically was related to the language being close to the machine representation).
However, the huge exception is CommonLisp. Lisp is the most powerful language that is, representing the classical extreme choice for expressive power instead of efficient implementation. ...
For new projects in which one is free to choose the language, one should choose Lisp."
Here's one where Visual Basic .NET beats C:
http://www.fourmilab.ch/fourmilog/archives/2005-09/000588.html
Java beating C:
http://openmap.bbn.com/~kanderso/performance/java/index.html
Of course C is an extremely primitive language. It is possible to (on current hardware) write a C program to best most other systems. But, a proper program is much more difficult!
Microsoft LOSES market share if you install something else over top?
I would beg to differ -- all machines are still counted as Windows(tm) machines. This allows demonstration of the ABSOLUTE and CRUSHING numerical superiority of Windows.
Must be, the number of attempts per second makes no sense. I get a COMPLETELY different answer based on similar "data" from his article:
Even if the password is "trivially" generated:
symbol word symbol word symbol: eg. _orange*bear#
would require 10,000 attempts per second to crack under these circumstances (roughly). Assuming that the defender uses ONLY this exact pattern:
The symbol characters add 4 bits each, or 1000 multiplier for 3 (very rough). Choose two words from the dictionary, say limited to 10,000 choices each (the author states that it could be chosen from a set of 50,000). 100,000,000,000 possibilities to run through, If 100 days given, 1,000,000,000 have to examined per day, 86,400 seconds per day, or just over 10,000/second. Contrast to 65/minute.
And that assumes a VERY limited password.
I should download and review this, but I really don't care enough.
But I will try to guess where that number could come from:
Two words together, nothing else. 100,000,000 combinations, 1,000,000 per day, or just over 10 per second.
A single dictionary word, followed by a number. 100,000 combinations. Finally, our attack rate can be 1 per second.
"In the bedroom". No. But what I need is POSIX compatible systems. I am happy enough with Linux, although I prefer Solaris.
As long as the operating environment supports what I need, I am willing to entertain it. Windows(tm)? Last time I looked, Microsoft was busy deprecating getcwd(). Apparently, ISO C++ insists on _getcwd(). I didn't know that Microsoft was particularly ISO compliant either (or maybe I'm stupid). MSVC (my version - 7), doesn't even support stdint.h. I guess I COULD rewrite a crapload of code, but it isn't happening.
But then, I guess my needs aren't "typical". I like OpenOffice.org presenter, with the current slide/next slide/clock/notes view. If I present and have to use the Microsoft product, I need to remember to find a suitable clock. But then, I guess I am not typical. I need a platform that integrates into Unix platforms, and supports NIS and NFS. But then, I guess I am not typical.
I have tried Windows and other Microsoft products. They STILL aren't suitable for me. But I am ever hopeful...
"No one cares that it may be 5% slower at foo task when you're running it on hardware that is 500% faster than the gear you replaced."
Wouldn't that mean that the OS itself is now taking in excess of at least 80% of your system resources?
(for the humour impaired -- I'm making a literal math joke)
Linux Lost against Windows 15 years ago?
Would be 1994. BEFORE WINDOWS 95.
And what, were you using then? Windows 3.11, I presume. In which case, Linux 1.0 (released March, 1994) offered a 32 bit system with memory protection and preemption.
So, tell me, did you ENJOY running EMM386.EXE?
I guess you could have been running Windows NT 3.1 (or 3.5, released September 1994). That also offered a 32 bit system with memory protection and preemption. It also cost $280 back then. Linux? Free.
Stop trying to rewrite history.
OpenOffice.org is "amazing". I have no freaking clue what Microsoft Office(tm) does, but I do know some parts of it having been forced into using those parts.
The parts that are important to me are:
Giving presentations: OpenOffice.org has an amazing control panel that shows current/next slide, time, and speakers notes on a separate monitor. What a godsend for presenters!
Editing PDFs: OpenOffice.org can do it.
Saving PDFs: OpenOffice.org can do it. I believe that MS Word(tm) can finally do this.
Can see fonts in menu: I believe later versions of MS Word(tm) copied this feature.
So the point of making it superior isn't actually true (Linux WAS ALSO clearly superior to Windows 3.1, 3.11 WfW, Windows 95 and Windows 98). "They" may not come, anyway.
I used to think that I should evangelize (when I was younger, and not as wise) -- now I just say "You will know if you want to switch, and why. Do whatever makes you happy."
It a fallacy.
Assume an exploit on a platform, and that the exploit can spread. It will spread on systems of the same platform, and it doesn't matter WHAT the percentages are. If an exploit hits (say) 70% of Windows(tm) boxes, it will hit 70%. Whether the Windows boxes are 90% of the population, or 30%.
But, if Windows has 90%, it has numerically greater numbers, which is of desire (for ddos attacks, etc.). Supposedly, anyway (we'll be back to that in a moment) so greater number of blackhats look for exploits on that platform. They don't because Mac, Linux, Solaris, AIX, HP/UX, BSD etc. have such low figures that it wouldn't be worth it!
Ok, this makes sense for credit card data, and the pride of pwning a large number of systems. Doesn't make as much sense for bandwidth. Most of the boxes that have access to the highest bandwidth run Unix (Solaris, AIX) or Linux.
So either the Unix/Linux boxes were not considered valuable targets, or it was beyond the blackhats expertise. XP was shooting fish in a barrel; Vista is more promising on a security level. Note that social engineering "trojans" don't work nearly as effectively on those high-bandwidth systems (typically headless, with professional admins).
In answer to your question, in the 30/30/30 world, the Windows (at XP level) would still have the largest number of infections. Mac would be far less (simply no open ports), but there is a possibility of a Mac trojan (and TFA relates to that exactly). Linux? Also has an interesting infection vector for trojans based on desktop files. Neither Mac nor Linux would allow an easy automatic virus, although Linux would be /slightly/ easier. Once in, the Mac box is easier to exploit further.
With Vista, the automatic virus propagation is mitigated, because, even if/when the box is "exploited" via a service, the exploit code cannot reliably do anything (at least this month).
As usual, YMMV.
Document Automation?
Sure, TeX will allow you to "program" documents, but I doubt that it what you meant. TeX sources files "written" as standard ASCII files. Which, in a Unix environment, means that the standard environment applies. You can use awk/sed/perl etc. to "automate" your documents.
Word(tm) and OpenOffice.org (because it wants to be "competitive" with Word) both incorporate their OWN programming language. About the only good that I have seen come of that idea is that Word could be used as an entry-level programming environment for Windows(tm). That this idea could be (seriously) floated boggles my mind.
Given that the ENTIRE purpose of the Unix environment is exactly that -- file and document automation, and drills down to that end with laser focus, duplicating that functionality in the "word processor" is stupid. The purpose of TeX is to apply that same laser focus to producing those "immaculate looking documents".
Separating functionality allows the parts themselves to be clearer and, as a corollary, more defect free.
As to standards -- the POSIX environment is the standard. TeX uses ASCII which is the standard. The TeX processor itself is arguably stable enough over 30 years to be considered standard. Which Word standard are you referring to? The one ratified last year I presume.
What is a "Word form system"? I am presuming you are either making a form in Word, which wouldn't be a system, or (more likely) using Word to fill out forms and file the results. In other words, using a Word Processor as a data entry system. Isn't that a bit like mowing a lawn with a shaving razor? To Microsoft's credit, once someone gets a thing like that working, it will be VERY painful to port to anything else (moving is simply impossible).
"We could easily construct a wishlist or consumer apps for Linux. My list would include Adobe's Photoshop Elements and Premiere Elements, as well as their professional graphics products (Dreamweaver, Illustrator, InDesign, Photoshop). I'd also add Intuit's Quicken and TurboTax, Roxio Creator or Toast, a Linux equivalent of WinDVD or CyberDVD, drawing programs like Visio and CorelDraw, a painting program, a website creation program like RapidWeaver or Freeway Express, and much more. There's also a very long list of educational titles and game titles that are almost entirely absent from Linux."
I don't know what Photoshop Elements is. I use gnucash for financial management. gnomebaker for making CDs. WinDVD? I don't know -- my linux laptop came with something that plays DVDs. I use dia for drawing, and kolourpaint for painting. I haven't bothered with a website creation program. As for educational and games - generally, I don't bother. The kids use flash games on the web, as well as educational resources.
All of this software came with my current laptop "integrated" (pre-installed, or manufacturer supported download). It's an Acer Aspire One, running the horribly named "Linpus" system. Adobe Reader and Flash pre-installed as well.
I did install an mplayer codec pack to allow (media-du-jour) to play.
Of course, having the list is of use to me as well; I need to know the Windows equivalents occasionally.
Certainly -
Adobe - Flash and Reader.
Microsoft - Via Novell, Moonlight.
Oracle/SUN - Java, OpenOffice.org.
AMD/ATI, nVidia.
As to going into a store and buying this stuff -- very unlikely to happen. 25 million? Probably already done with StarOffice and MySQL support.
Trying Linux since 1992, eh?
You are complaining about Printing, Networking, Laptop features.
What were you using in 1992? Windows 3.11, WfW? Or, did you use a Mac? Maybe OS/2? Presuming you were using "standard" architectures, because that was all Linux ran on back then...
Back then, Linux was far and away the technical lead. Except, of course, when compared to OS/2, or SCO. (My "rant" apologies if you were on one of those platforms).
Linux offered REAL processes and preemption, and REAL isolation. It was able to support real servers, and REAL applications. Windows? Mac? Not so much. Simply holding down the mouse button could crash networks with MacOS.
Linux proceeded to develop, as did Windows(tm) and Mac(tm). Apple eventually introduced OS X, and Microsoft the NT based kernels. Bringing their stuff to parity.
As a Unix (Linux) user, I didn't really bother with USING Windows(tm). It didn't offer anything and wasn't compatible with the applications I wanted to run. Finally, Windows(tm) caught up, as did Apple with OS X. Still, there are missing pieces (NIS, NFS others). At least Apple owns CUPS now.
As to printing: (to give a specific)
The Unix model for the past twenty years has been to have the applications generate Postscript. That will go to a standard LPD, and may be sent to a different machine. If the printer being used is "inferior", GS may be used to generate a raster for it. This is, of course, isolated from the application.
CUPS adds printer publishing (don't need to define the print queues anymore).
And that's all there is to it.
Windows(tm) breaks this model. GDI is used to rasterize and drive the printer. Completely different, and in my opinion, wrong and broken. CUPS does try to accommodate the Windows(tm) model, but it is, at best, a poor emulation. Of a broken system, so it doesn't matter to me anyway.
You can try to argue that a GDI based model is superior to a Postscript model -- but Postscript has far better resolution and device independence. It IS the standard for Unix printing. Which is NOT Windows(tm) printing.
Yes, I am one of those crusty "Here's a Nickel, Kid, Get yourself a better computer" guys.
http://tomayko.com/writings/that-dilbert-cartoon
Yes, Windows(tm) and the Mac OS have (for the most part) finally caught up. But they don't really offer much extra; except for some specific applications.
If you NEED one of those applications you would be better served with the system that actually supports it. Personally? Since I am an old-hat Unix/Linux user, I don't have any such applications.
A day playing with it?
Um... no. If I buy a copy, I cannot return it. And I will not "pirate" it. So, it's not happening. If it doesn't run on Solaris and Linux, I can't use it. So, it's not happening.
Which means I use OpenOffice.org. With menus. If I have to use MS Office(tm) I, of course, go nuts. Never had the "training" or familiarity.
So, I say, "stuff it". If it isn't OpenOffice.org compatible, take a hike.
Just sayin' We're not all Microsoft users, ok?
Typical use case:
I am currently engaged with a client. For a number of reasons, we had to accept "unlimited liability" in the case of information leakage (and that could easily run into the millions).
If I have a VPN connection linked up to the internal network, I cannot tolerate other external connections. Whether or not "malware" is transmitted internally is not my concern (and it probably isn't an issue; I use neither Windows(tm) or OS X).
Further, I don't store "sensitive" client data on my machine; other data is stored encrypted. Even if I am physically at the client site, there is a separation with network cables -- some have external access, and some are internal. To bypass this, I guess I could connect TWO cables, but my machine only has a single network port preventing this from occurring accidentally.
This VPN behavior is useful to those of us who need an "air-gap" or something arguably similar. Yes, I use VPNs like this, and I guess you could call me paranoid.
No.
It would SEPARATE hardware innovation from software innovation. Let the hardware companies do their own thing, and supply specifications.
What may happen is that time to market increases. Basically, the new hardware has to be seeded (or purchased) by software developers and support organizations before general adoption.
Large software houses would be more conservative with support, while smaller, less risk averse houses would offer earlier support.
In other words, Microsoft Windows(tm) would no longer be the FIRST platform to be supported. As a corollary, it would not be the ONLY platform supported.
But, there will be vendors developing drivers into the Microsoft ecosystem as well.
Cost of hardware should also be reduced. Users of BSD, Linux, and other "free" platforms won't need to pay for the price of Windows(tm) drivers.
Commercial driver support will be possible, and not just as contracting to hardware companies (which is the current situation).
Since hardware specification would be available, low level software innovation will again be possible.
It would not stifle innovation. Indeed, the opposite would happen.
Canon sells cameras, not software. The software is added as an "extra". Since the software IS supplied, Canon tries to support it, and limits the supported platforms (trying to control costs).
If Canon were liable for the "licensed, not sold, software", they would have to do something about the software. Either improve its quality (and cost), or, more simply, not supply software. The camera would then be hardware conforming to a specification. Look at modems, printers, and other peripherals in the pre-Windows 9x era. The devices came with clear technical interface manuals, or compliance to standards. Others were expected to write the "drivers" to make the devices useful. Sometimes the vendor would supply software, but more as an example, and almost always with source code (the example should be useful).
[Sidenote: This practice was even extended into the early PC computers. Apple published source for the Apple ][ ROM, KIM-1, SYM-1 had the source supplied, and even the IBM PC had the BIOS listing]
This would again allow platform variation. The smaller platform suppliers would be able to support new peripherals FASTER than the larger vendors, thus, in my opinion, (re)balancing the software landscape.
But...
My internet is already balkanized. As much as possible. Because I don't trust you. It's MY network, and I'll allow YOUR network access if and when I trust your network.
So far, my trust network is only to two other networks. I know the admins personally. Quid Pro Quo, one of the admins is my outbound mail relay.
Yes, I'm probably paranoid. But really, everyone else gets minimal access and services.
"Replacement for Windows"?
I think you will find that quite a few people were already on the "cutting edge". Using OS/2 and such. Windows wasn't even an "also-ran", it was, at that time, a complete non-starter (1991-1996).
Linux is not the next thing after Windows -- by any measure it is a contemporary. The question that many have is "why doesn't Microsoft make Windows more compatible"?
It helps to revise history (from the winners perspective), of course. But, back in the early '90s, Linux wasn't in competition with Windows (Windows 3.1, WfW? Give me a break). It was in competition with SCO (and, to a lesser extent, BSD).
Windows was NOT even remotely the target -- it had already been surpassed (technology-wise). You could build a (reasonably reliable) server based on Linux 0.99. Try that with Windows 3.1 or Mac OS of the day!
Linux allowed easier experimentation with OSs than Minix, and BSD was caught up in some disputes. SCO was too expensive.
Wrong. The right of property gives the right to reproduce. After all, you can do anything you want with your property.
This right is limited by Copyright. ("takes it away")
Copyright is NOT the start of this reasoning -- it falls under.
The creator certainly has the right to not share. If the creator desired absolute control, she may decide not to share, or to apply Trade Secret or other (contract) provisions. But, if the work is shared, it becomes "open" in the sense that the creator is not deprived of her ownership by the fact of other ownership. (which distinguishes "Copyright violation" from "theft").
What I find coolest is the "+5" on your post.
And why not?
Let's examine some history.
In Unix (Linux), there is a rigid placement for code -- /bin and /sbin contain the OS utilities, /lib for neeed libraries, and /boot contains the OS itself (/kernel for Solaris, but the idea is the same). /usr/bin, /usr/sbin, /usr/lib et al contain OS utilities NOT NEED FOR BOOTING (/usr can be network mounted). /etc for OS configuration. /usr/local and-or /opt for site specific programs. (Applications not supported by the OS vendor directly, or, in the case of Solaris, by a different group)
Easy to upgrade the OS. Replace the OS parts, and leave /usr/local, /opt and the /home (/export/home) directories alone. No particular complications. Everyone administering a box knows the score, and we don't have any problems. Of course, /etc changes are then the worst, but apps can be encouraged to go with /usr/local/etc for global configuration (this is rarely done, it turns out to not be a practical necessity).
As to what is installed? Different OSs use different conventions (obviously, and Microsoft uses their own). yum list in Fedora (for example) will give the list of packages, libraries and utilities currently installed via the vendor installer/updater (2373 installed packages on my netbook I am using to type this). It does behoove Microsoft to keep track of all of their own packages, beta to beta and to release (it would be part of bug tracking).
Given the experience in this area gained from 40 years of use, I would imagine the Windows Update procedure to be as finely honed, especially given the new security concerns of Microsoft in relation to Windows.
So it doesn't make sense to me that a beta to beta update isn't supported. This must be strictly a "scope limiting" move; but I don't think it reflects well on Microsoft and its position as an OS vendor.
"There is no inalienable right to use copyrighted material"
There sure is. The right of property. I assume that you are talking about US law, given the article to which you are commenting.
The Bill of Rights prohibits the federal government from depriving any person of life, liberty, or property, without due process of law.
A CD, book or DVD is simply property and you can use it however you want.
Of course, you are talking about Copyright, which, in this case refers to the right to make a copy of the property in question. Since the Bill of Rights trumps Copyright, Copyright cannot restrict what you do with it, outside of the (hopefully narrow) proscriptions specified and justified in the Copyright Act.
So, there is an inalienable right. (Well, technically, I guess you COULD argue that the Bill of Rights is not inalienable, but it should be so considered in the US).
What we are talking about is then the scope (or restrictions) placed on Copyright, and the reactions of Companies to overly broad interpretations of Copyright as promoted by cartels and not creators.
"Double the pipeline length...directly led to lower instructions per cycle"
Wouldn't twice as many instructions be worked on per cycle? Shouldn't this result in a an increase in instructions per cycle, for "properly" written code?
"Things usually go like this" -- Maybe, but I have used computers for a LONG time. It went more like this:
(Note that ALL details have happened to me, as listed, but not necessarily in the order presented. Because of my Windows XP "experience" I refuse to buy Windows Vista.).
Windows Fanboy: Windows XP is great, it does anything you need it to do!
Greybeard User: That's nice. The last time I tried it was Windows 98SE, and that version couldn't really walk and chew gum at the same time without dying. Maybe I'll try again.
Greybeard User: I tried to install in (after spending $200 on it) in order to run a TV tuner recorder, but it doesn't recognize the CD it just installed from.
WF: It needs drivers.
GU: Right, but it doesn't recognize the Network, or USB, and the drivers are too big to fit on a floppy...
WF: Just "slipstream" them.
GU: What? I don't HAVE another computer running Windows XP.
WF: You must be a Unix Loser!
GU: I think I'll try this "Fedora Core" thing. Hey! it works.
GU: [after many months pass]. Ok, I have this "Windows XP" running on a VMware instance. Why don't I have a NIS option for login? Don't ALL Operating Systems support that?
WF: You must be a Unix Loser. I don't know what that is!
GU: The command line sucks. Why can't I have a POSIX shell like Bash?
WF: It does not suck, you tool! And, you can use Cygwin to run Bash.
GU: And, it doesn't run the programs I have been using. Nothing works!
WF: Use Cygwin, Duh!
GU: What is this Cygwin you keep talking about?
WF: It lets you run Unix programs on Windows. RTFM noob.
GU: What manual? I bought Windows XP, so it didn't come with a paper manual and if you don't know what you are looking for, the on-line help isn't very helpful.
WF: Umm...
GU: If Windows is so great and does everything I need it to, why is something like Cygwin even necessary? And, even WITH Cygwin, I still can't use NIS and standard logins. And, why is Cygwin fork function so damn slow? My programs run like a tortoise in molasses.
WF: Because of tools like you! You are just a Unix Loser! Windows is great so just shut your hole!
Windows: where getcwd is deprecated for _getcwd
"Windows 7 has proven to be the most stable Windows release for a good decade."
Really? Isn't Windows 7 only in Beta? So, it's not a release yet, is it?
Isn't your comment, taken as a whole, a rather strong indictment of Microsoft OSs? Why would you consider subjecting yourself to that again and again?
Just wondering.