OT: RIght

Prompt, too.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

JP > 2011 Japan Prizes Awarded to : Dr. Dennis Ritchie and JP > Dr. Ken Thompson for the development of UNIX JP > =A0

JP >

JP > And the C programmiong language. JP > Well deserved!

TW > Prompt, too.

LOL

C set the world back by decades. The Nobel prize will go to whoever gest us out of this mess.

John

C was a good solution of its time for writing operating systems. I preferred some of the syntactic sugar of BCPL notably tagged $(.

Industry chose to go for weakly typed badly behaved languages that were quick to compile and suited hackers. It doesn't make much sense today to use tools that are so prone to facilitating bad coding, but we are where we are. CPU cycles are cheap and getting cheaper whilst human cycles are expensive and well trained ones getting rarer.

Computer science and military went towards strongly typed and harder to use languages like Algol, Modula and Ada. It depends what you want - if time to market is everything then the hackers win every time and customers suffer the consequences. HMRC appear to have mislaid £1$b or so of National Insurance comtributions in the UK according to a recent computer audit - all due to "improved" software.

Not clear to me that the second generation languages will ever be sorted out. It will need a paradigm shift equivalent to the jump between autocode and true HLL compilers to bridge the gap between demand for new reliable software and ability to supply it.

Regards, Martin Brown

On a sunny day (Wed, 26 Jan 2011 06:57:27 -0800) it happened John Larkin wrote in :

But I wrote that especially to wake you up :-) I am happy with C, and C libraries, and Unix. So is Jobs, Android, Linux, universities. As long as I can write stuff like this is an evening what more could I want? ftp://panteltje.com/pub/spec1.gif Just got a nice scintillation crystal. C or an OS is not a purpose or goal in itself, it is a tool. If you do not know how to use a tool it is your problem. There may be other tools that can do the same (your BA-SICK for example), but that is irrelevant. C and Unix has a much wider base and more powerful libraries available. I only used: grml: ~ # whereis xscpc xscpc: /usr/local/bin/xscpc grml: ~ # ldd /usr/local/bin/xscpc linux-gate.so.1 => (0xffffe000) libpthread.so.0 => /lib/tls/libpthread.so.0 (0xb7eef000) libforms.so.1 => /usr/X11R6/lib/libforms.so.1 (0xb7e5e000) libm.so.6 => /lib/tls/libm.so.6 (0xb7e39000) libXpm.so.4 => /usr/X11R6/lib/libXpm.so.4 (0xb7e2b000) libc.so.6 => /lib/tls/libc.so.6 (0xb7cf9000) /lib/ld-linux.so.2 (0xb7f18000) libXext.so.6 => /usr/X11R6/lib/libXext.so.6 (0xb7ceb000) libX11.so.6 => /usr/X11R6/lib/libX11.so.6 (0xb7c2c000) libdl.so.2 => /lib/tls/libdl.so.2 (0xb7c28000)

And takes no space either on most system, as those libs are already present grml: ~ # l /usr/local/bin/xscpc

-rwxr-xr-x 1 root staff 27892 2011-01-26 16:26 /usr/local/bin/xscpc*

< 28 kB.

Wonder how much the MS bloat would need, probably 400 MB for install to begin with LOL.

C is simple, here is your first lesson, compile it with gcc -o test test.c then run it like this: ./test

/* test.c */ #include #include int main(int argc, char **argv) { int we_have_fun = 1;

while(we_have_fun) { fprintf(stdout, "whoopy\n"); } }

On a sunny day (Wed, 26 Jan 2011 15:34:25 +0000) it happened Martin Brown wrote in :

Bit short sighted. If you are a moron then your coding will look like that. If you give a monkey a typewriter it will make a LOT of errors.

You can give a monkey an automated machine that uses words from a dictionary, and it will type nonsense without any spelling errors.

You can give a monkey an automatic translation program, and it will happily translate Chinese into English by hitting a button, but it will still look weird.

You can give a monkey a personal human servant to do the work right.

C is a tool, if you do not know how to use it, or are too fragmented in your attention not to make silly mistakes, should you, with a mind set like that, be allowed to drive a car? Do you know how many people die in traffic accidents each year?

Does that make cars bad, are you waiting for that project were all the cars drive behind each other at a fixed distance automatically? STILL somebody will accidently open a door and fall out. Oh, but we can automate that door too.

Get a life,.

It is precisely that that people have become weasels and no space exploration is done, and all chemicals are considered dangerous, Did you light a match lately? Did you know it can burn down your place?

With humanity going that road it will go the way of the dinos, and mosquitos will rule, politically incorrect, the earth.

Most programming could some day be done with non-procedural languages, vaguely LabView sorts of things. The paths through the maze of a procedural program - much less a multithreaded one - is just more states than most programmers can manage. It's another order of infinity.

Ada is better than C because it removes a lot of hazards and allows a lot of checking... if you want to do a lot of checking. But most hackers don't have the patience to work in Ada. And it's still procedural.

Probably the most successful programming language so far is COBOL.

John

Some people are. Some people can write good C. But big projects, coded by armies of mere mortals in C, tend to be huge inefficient messes, and often total failures. How many bugs and hazards has Windows had so far? Hundreds of thousands? A bug every 10 lines of code? I know of a bunch of gig projects, millions-of-lines things, that are train wrecks.

Decent resolution for a scintillator. But single-evening projects by one smart programmer aren't the real problem these days. I could have done something similar in PoweerBasic, or even assembler, but small-scale methodologies often don't scale.

John

There's an argument that says this is a good "market" solution. And it's not clear that all the foofraw of strong typing does much of a practical nature for us. Even 'C' compilers can get pretty persnickety about typing if you use the right options.

Yarg. Sounds more like an organizational psychology problem than a technology problem to me.

The political economy of it all probably means that the present equilibrium will last a very long time. IMO, the paradigm shift happened decades ago - with LISP.

Hard to write a bootloader in LISP, though.

-- Les Cargill

Hey, weakly-typed is better than the no-typing-whatsoever that assembly gives you. :-)

But I do think you're right -- the kind of guys who develop new technologies tend to be the hacker/super-programmer types, and they're going to build tools that emphasize speed and agility over safety; as a particular technology becomes more mainstream, there's a natural evolution of better safeguarding to protect the now-more-mainstream users. :-)

That's not always true, though; look at something like OrCAD: The old version for DOS, developed something pushing two decades ago now, is considerably less buggy and faster than the Windows version, despite being developed with with far less capable tools. So I'd argue the main difference is that today the

*mean quality level of programmers* is rather less than it was some decades back, and *modern tools can sometimes compensate for this* .... but you're still FAR better off *just hiring really good programmers in the first place!* (...and having them use modern tools is often a good idea too, when possible...)

A visit to thedailywtf.com is always useful to see just how bad some programmers are!

The other big change is trying to figure out how to efficiently/easily support multi-threaded programming: We've largely hit some upper limits on CPU speeds, so for the past decade now it's all been about multiple cores, yet today all the popular languages still leave it up to the programmer to figure out how to parallelize their code.

---Joel

On a sunny day (Wed, 26 Jan 2011 08:21:31 -0800) it happened John Larkin wrote in :

Microsoft Windows is written in C++ I believe. C++ is a language deficiency, like a speech problem. This man is responsible for that:

Maybe bs stands for B*llSh*t

Hundreds of thousands? A bug every 10 lines of code? I know of a

Yes, uses a very nice crystal I bought on ebay. Very sensitive too.

Scale to what? The world if full of small embedded stuff, soon all interconnected with IPV6.

A poor craftsman blames his tools.

If you want to commoditize s/w development, then C isn't the way to go. Better to stick with a strongly typed, restrictive language and development platform. But the fixed costs to set up and manage such a development environment are high. So there's a drive to make everything a big project so as to recover these costs.

--
Paul Hovnanian  paul@hovnanian.com
----------------------------------------------------------------------
Have gnu, will travel.

Which part of it?

Until about Windows Vista, Microsoft themselves didn't write the Kernel using C++ (just regular C) -- although some 3rd party device drivers were already being written in C++ about the time Win 2K/XP came out.

For user mode apps, early on C++ was popular, but in the later days of XP and all of Vista and Win7 the .Net languages have become well-represented too.

So what do you think of C#, then? :-)

---Joel

On a sunny day (Wed, 26 Jan 2011 10:53:13 -0800) it happened "Joel Koltner" wrote in :

Well, I used Visual Studio for my work, and that was ++.

I am not a MS widows expert, and sure in the beginning maybe there was not even C++ (win3.1 times perhaps). I have read they kept porting 16 bit device drivers and stuff, maybe even just ported the executable code, I dunno. I *DO* know that most MS windows applications usually ask for a huge 'install space', crash on a regular basis, and provide little functionality for the code size, be it the executable itself or the DLLs. So bloated and crap. I got so fed up with that stuff that I burned my xp disk and removed the crap and repartitioned the disk for reiserfs after the second re-install of xp (re-install an OS? insane), and it finally had it working (took me whole Saturday, just staring at the stupid MS commercials while xp was 'installing', and then I switched it on Monday morning and it told me my mouse was a MS mouse, and the cursor was out of control, all aver the screen. I do not even HAVE a MS mouse, it is a Logitech mouse. So that did it for me, MS foundation classes, whatever they used, and closed source, nothing you can fix, lookup, or improve. I burned the xp disk and made a video from that, it is available for 110$, you have to accept a non-disclose paragraph, put a sticker on your PC with a special number, I may come busting into your place it should be there, else you have to pay thousands of damages to me. So, anyways, I mean that xp could not even do anything, I tried their video editor, it was slower than a sick snail.

I have never tried C#, I have no idea what it is, what it does, and why I should need it.

grml: ~ # uname -a Linux grml 2.6.21 #1 Fri Jul 17 14:57:16 CEST 2009 i686 GNU/Linux

LTspice runs on Linux in wine (windows emulator), for all other things I either wrote the application myself or there is one with source code available. From Open Office (LibreOffice) to cups (printing), from Gimp to imagemagick (pictures), from dvdauthor to transcode (optical disks), any communication related program you can imagine, from cryptology to FPGA tools, compilers, assemblers, also for many micros, cross platform, I could go on for hours, WHY would anyone need MS bloat.

runs on a 10 year old mobo with that kernel, same PC does the home automation, the video recording, security webcams in H264, satellite recording, ftp server, mail server, is part of DNS (named), does audio (2 sound cards), all those things at the same time, plus web browsing, flash 10, xpdf pdf viewers open and it is not even fully loaded;

top - 20:20:42 up 7 days, 1:28, 10 users, load average: 0.35, 0.41, 0.44 Tasks: 114 total, 6 running, 108 sleeping, 0 stopped, 0 zombie Cpu(s): 1.0% us, 5.0% sy, 31.7% ni, 58.1% id, 0.0% wa, 1.7% hi, 2.6% si Mem: 386028k total, 373256k used, 12772k free, 41564k buffers Swap: 499992k total, 149912k used, 350080k free, 135956k cached

All that in 380 MB of memory.

Sure MS and industry work together to sell you ever more powerful hardware, the hardware is needed because of the OS bloat, and application bloat.

Soon you will need a 6 core to write an email 'hello world'.

So, I rather stay with this system. If I need anything that I cannot find, or do not like the available stuff, or it simply does not exist, then I write it:

Use open source, and give some back to the pool.

Just as a point of reference, MFC actually does come with its source code, as do the C run-time libraries (MSVCRT). Or at least they did last time I had to use them, some years back...

You're correct, of course, that most of Windows doesn't, though... unless you happen to work for one the few large companies that has a source license agreement with Microsquish.

Yeah, but if you use WINE you're still condoning the use of the Win32 API standard, you know? :-)

Well, it's not like modern Linux distributions recommend any less (hardware power-wise) than Windows does. Windows XP is now over a decade old, after all!

---Joel

I am so impressed. NOT!

That is *huge* for such a minimalist program. The smallest that any of my tools can make it on a PC compiled with a HLL is JPI/Topspeed Modula2. The source code is shorter too and a lot more explicit.

MODULE HelloWorld;

FROM IO IMPORT WrStr, WrLn;

BEGIN LOOP WrStr("Hello World");WrLn; END END HelloWorld.

19.5kB DOS executable with no optimisation (self contained). Compresses to a 2,644 bytes EXE with PKLITE.

ISTR modern ISO M2 version requires s/IO/InOut/ s/Wr/Write/

Not bad for a compiler that dates back to the mid 1980's.

A modern day compiler would warn that the program never terminates. Any compilers and smart linkers that can beat 2,644 bytes DOS EXE? (for "Hello World")

I wonder how big PowerBasic makes it?

with LOL.

You assume that all compilers on PCs are written by MickeySoft. That is incorrect - there are other high quality compiler vendors around. Even the Intel compilers have some interesting features in their profile directed optimisation and deeper hardware access.

How typical of a C hacker to write a trivial program that never terminates. Obviously never heard of the halting problem.

Regards, Martin Brown

On a sunny day (Wed, 26 Jan 2011 20:13:05 +0000) it happened Martin Brown wrote in :

28 kB for the gamma spectrum display program is really small. It has also a complete terminal emulator build in, and control functions for the PMT. Look at the C source (old version):

I dunno, I can probably write that in x86 asm in a few bytes calling BIOS.

Now please write the spectrograph in yo uown language, and then we compare sizes. It is easy, you have the C source as staring point. Watch out it is multi-threaded.

with LOL.

Idiot, even a newby knows about ctrl C. Obviously your brain halted.

Bye

On a sunny day (Wed, 26 Jan 2011 12:05:02 -0800) it happened "Joel Koltner" wrote in :

Yes, it is up to LT to port it to Unix. But I am very happy with it anyways.

Maybe if it is Novel, or 'Open Suse' or whatever, they are in bed with MS. Or RatHead, they are money making machines.

But look at this, this is what I run now (old version):

There are many < 1GB solid state (SDcard) based Linux 'distros' too, I have Puppee for the eeePC for example. The whole eeePC is a 4GB FLASH chip with 1 G of RAM.

I have an old Linux on a 128 MB USB memory stick too somewhere.

To big, long-term, multi-programmer projects.

John