Thread: Is there a standard function?

Originally Posted by iMalc

Okay final version. Pulled out all the stops now... Only two divs or mods in the whole thing. I even got rid of two gotos! Time to retire this one:

Code:

int iMalcsUltimateFormat(long snum, char s[])
{
    char *ps = s;
    unsigned long num1 = snum, num2, num3, div;
    if (snum < 0) {
        *ps++ = '-';
        num1 = -snum;
    }
    if (num1 < 10000) {
        if (num1 < 10) goto L1;
        if (num1 < 100) goto L2;
        if (num1 < 1000) goto L3;
    } else {
        num2 = num1 / 10000;
        num1 -= num2 * 10000;
        if (num2 < 10000) {
            if (num2 < 10) goto L5;
            if (num2 < 100) goto L6;
            if (num2 < 1000) goto L7;
        } else {
            num3 = num2 / 10000;
            num2 -= num3 * 10000;
            if (num3 >= 10) {
                *ps++ = '0' + (char)(div = (num3*6554UL)>>16); num3 -= div*10;
                *ps++ = ',';
            }
            *ps++ = '0' + (char)(num3);
        }
        *ps++ = '0' + (char)(div = (num2*8389UL)>>23); num2 -= div*1000;
L7:
        *ps++ = '0' + (char)(div = (num2*5243UL)>>19); num2 -= div*100;
        *ps++ = ',';
L6:
        *ps++ = '0' + (char)(div = (num2*6554UL)>>16); num2 -= div*10;
L5:
        *ps++ = '0' + (char)(num2);
    }
    *ps++ = '0' + (char)(div = (num1*8389UL)>>23); num1 -= div*1000;
    *ps++ = ',';
L3:
    *ps++ = '0' + (char)(div = (num1*5243UL)>>19); num1 -= div*100;
L2:
    *ps++ = '0' + (char)(div = (num1*6554UL)>>16); num1 -= div*10;
L1:
    *ps++ = '0' + (char)(num1);
    *ps = '\0';
    return ps - s;
}

Very good, dude. :-)

this evening, back home, I'll give a better look at that.

I'm actually unable to stay longer.

Bye for now.

>_<

So I'm the only one using `log2', `sin', polar coordinates, and a massive array to do this?

FINE!

*thplblblbltlttlbltl*

Soma > off to pout

Originally Posted by phantomotap

>_<

So I'm the only one using `log2', `sin', polar coordinates, and a massive array to do this?

FINE!

*thplblblbltlttlbltl*

Soma > off to pout

What are you talking about phantomotap?
:-)

@Elysia:

You had to add an underline in front of a function name -- GET OVER IT ALREADY, and quit yer damn whining.

Originally Posted by phantomotap

>_<

So I'm the only one using `log2', `sin', polar coordinates, and a massive array to do this?

FINE!

*thplblblbltlttlbltl*

Soma > off to pout

Did Frkton time your program on his system, Soma?

Code:

But you didn't, and therefore it complained about it being a POSIX function. All I did was point that out.
Can we agree to stop this now? If you say yes, I'll say yes.

@Elysia:
I didn't add the underline, because my compilers either don't allow and/or require it. That was the only ultoa() function that was portable between my two compilers.

You did not just "point it out", you launched into your same old tirade about my compilers, and I'm tired of you whining about them. I've spent several hundred dollars on these compilers, and I have some large programs that don't work on any other compilers. I need to use these two compilers. You can .......... and moan about it as long as you like.

Originally Posted by Adak

Code:

But you didn't, and therefore it complained about it being a POSIX function. All I did was point that out.
Can we agree to stop this now? If you say yes, I'll say yes.

@Elysia:
I didn't add the underline, because my compilers either don't allow and/or require it. That was the only ultoa() function that was portable between my two compilers.

You did not just "point it out", you launched into your same old tirade about my compilers, and I'm tired of you whining about them. I've spent several hundred dollars on these compilers, and I have some large programs that don't work on any other compilers. I need to use these two compilers. You can .......... and moan about it as long as you like.

Ah! Ah! Ah! Ah! Ah! :-)
From my point of view you can use whatever you like, whenever you
think it is fitted for your pourposes, and to whatever extend you can imagine.
:-)

By the way, did you come with any turbo C/MSVSC/WTK idea about the
main task of this thread?
Let me know and enjoy.

@cyberfish:

if you can help me solve these compiling errors of your code
I can try it on my testpad:

Code:

 error #2048: Undeclared identifier 'BUF_LEN'.
 Missing prototype for 'memmove'.
*** Error code: 1 ***
Done.

Originally Posted by Adak

One program from each one, actually. I'm looking forward to seeing how Whitehorse's, Soma's, and iMalc's programs do in your testing.

I tried iMalc's code, and it looks quite efficient as I already posted. :-)
Don't know where are soma's and Whitehorse's so far, let me look for them.

Originally Posted by kermit

Code:

stop = clock();

    sign = false;
    int_format((int)(stop-start));
    printf("\n Elapsed ClickTime: %s",buffer);
    for (x=0; x < 15; x++)
        buffer[x] = ' \0';

Perhaps that is giving you some trouble?

Yes indeed. I managed to change that. Thanks.

Originally Posted by whiteflags

No this is basically a dick waving contest. Although AFAIK the only thing printf is guaranteed to do in response to locale changes is change the decimal point. Still important.

Anyway thank you for the contribution iMalc! This made for a fun profile. When I discovered my IDE had a profiler.

Anyway, in the interest of full disclosure, I unwound the recursion in my method and used another stack instead. It could probably be smaller. I did also profile my recursive version, but did not include it in this test.

I had to make significant changes to Adak's code to get it to compile. It probably isn't fair, but I had to do things like delete C++ comments and make sure his function actually printed output like the rest can.

In an upset, Adak's method is the slowest, with me trailing iMalc by a hair. If frktons wants to read the profile properly, I suggest this page. Profile output is not execution time -- it's something better. The actual execution time was something not pretty, like 800-900 seconds.

Because it may be pertinent, this is a debug build with no optimization from mingw (gcc 3.4.5), on a P4 2.6GHz architecture. I hope this is fair to everyone. I did my utmost.

Hi whiteflags.

I tried your code, but it displays million lines on the screen,
so it is not something I can test for performance. :-(

What should this test program shows? Remember I'm a beginner
and many things are obviously unknown to me.

I can analyse your code with my eyes, not with a profiler for the time
being, and get some insight/inspiration from it. But if you think it is
worth to try for performance, please post a suitable version.

Thanks

O_o

I'm sorry for the confusion. I did not post any source.

I only posted that comment to playfully mock the other posters.

*shrug*

Anyway. What can you really expect from a 1800 MiB array loosely indexed by repeatedly dividing `double' values?

Soma