TODO, FIXME, etc in Vim

Add this to .vimrc:

noremap <F6> <ESC>:vimgrep /\C\(TODO\\|FIXME\\|XXX\\|BUG\)/ ./**/*.c<CR>:copen<CR>:cc<CR><CR>

Small and fast build tools: SVMK, NINJA

There are many build tools. One of them tracks dependencies and build result output via special file with dependencies-tree description, another generates such file from universal format (Premake, Bakefile, CMake, etc.). There are 2 big classes of such tools: with special syntax in its' Makefiles, and with supporting of some general-purpose language (Waf, Scoons, etc.). Difference is that in the 2nd kind you can solve absolutely any tasks without writing special plugins (sometimes is impossible to extend "Makefile" in natural way, only with command line utilities)...

If you want very small and fast build tools (and not Make:) you can see next: Ninja and Tcl-based Svmk. Another Tcl-based make tools are here.

Ninja

Ninja is very simple and fast. It have not functions, if-else-for-while operators, nothing, only variables, rules... But it can be used to process automatically generated "Makefile". Example:

DMD = c:/dmd2/windows/bin/dmd.exe
LINK = c:/dmd2/windows/bin/link.exe

INC = -Ic:/dmd2/src/phobos -Ic:/dmd2/src/druntime/import

rule mkprj
    command = $DMD -I$PKGDIR $INC -of$out $in
    description = Compile D program $out

PKGDIR = misc/scripts
build $PKGDIR/email.exe: mkprj $PKGDIR/characterencodings.d $PKGDIR/email.d

Save it in build.ninja, call ninja misc/scripts/email.exe and you will get result: email.exe in misc/scripts directory. rule creates rule which has attribute command to call when it's applied. build set dependency: email.exe will be built from 2 .d files with mkprj rule by calling it command. Nothing else. Simple, fast. ninja is only ONE FILE: ninja.exe and is about 800Kb! If no target (misc/scripts/email.exe), then ninja will build it.

Svmk

It's Tcl-based make alternatives. So you have all power of Tcl (use Tclkit Shell!) in one little Tcl-script: svmk.tcl. Example of the same:

# vi:set ft=tcl:

set DHOME c:/dmd2/windows/bin
set DMD $DHOME/dmd.exe

proc dcomp {out in {dflags ""}} {
    global DMD
    if [llength $dflags] {
        system $DMD $dflags -of$out {*}$in
    } else {
        system $DMD -of$out {*}$in
    }
}

target email.exe {
    cd misc/scripts
    set SRC "characterencodings.d email.d"
    depend $SRC {
        dcomp email.exe $SRC
    }
}

We create dcomp procedure for compiling of D applications. Then create rule "email.exe" which is PHONY :) Common usage is target TARGET { depend DEPENDENCIES { commands... } }. You can see: we can use any Tcl commands, even download something via Net :) All what you need is: tclkitsh.exe: 740Kb, svmk.tcl: 10Kb!
To run:

tclkitsh.exe svmk.tcl email.exe

And sure you can (and have) to use original D build tool (if we are talking about D;) - dub.

Installation and trying D

First, download ZIP-archive with D compiler. Then unpack it into... for example c:/dmd2. You need directories: html and man (documentation), samples, src (for including), windows (binary, libraries) and README.TXT. Create somewhere test file, for example a.d:

import std.stdio;

int main(string[] args) {
    int[string] d;

    d["qwe"] = 12;
    d["asd"] = 13;
    d["zxc"] = 14;
    writef("%d\n", d["asd"]);

    int i = 10;
    typeof(i) j = 90;
    writef("%d", j);
    return 0;
}

Then create dmd.conf file (or sc.ini) with next content:

[Environment]
LIB="c:\dmd2\windows\lib"

DFLAGS="-Ic:\dmd2\src\phobos" "-Ic:\dmd2\src\druntime\import"
LINKCMD="c:\dmd2\windows\bin\link"
DDOCFILE=mysettings.ddoc

Now create Makefile:

DBIN=c:/dmd2/windows/bin
LINK=$(DBIN)/link
DMD=$(DBIN)/dmd
RM=rm -rf

DFLAGS=
LFLAGS=

SRC=a.d
OBJS=$(SRC:.d=.o)
EXES=$(SRC:.d=.exe)

.SUFFIXES: .d
.d.o:
 $(DMD) $(DFLAGS) -c $< -of$@


clean:
 $(RM) $(OBJS) $(EXES)

######################################################################
a: a.o
 $(LINK) $(LFLAGS) $<

Now you can compile first D program:

make a

and run it:

a.exe
13
90

Multiplication of big numbers represented as strings

This is the usual task in interview: solve multiplication of integer numbers presented as strings, for ex., "25" * "25" = "625". Length of these strings may be any (not 16 bits or 32 bits only).
Implementation will use school way of multiplication of numbers (sign is ignored), see:

   1856
   1856
   ----
  11136
  9280
14848
1856
=======
3444736

So, "1856" * "1856" = "3444736", we multiply each digit of second number on each digit of first one, shift result number by one to left, summarize them and don't forget about carry. Carry is of 2 kinds: carry of multiplication and carry of sum. Algorithm is:

• multiply last digit of 2nd number by last digit of 1st one
• write result (only low digit) in result string at LAST POSITION
• if resulting number is > 10, save carry (divided by 10)
• make the same with the 2nd digit of 2nd number, but instead of saving resulting digit, make summation: with carry of multiplication AND carry of sum
• don't forget to make shifting to left when go to next digit of 1nd number

So, each "saving" of result digit should make summation and there are not number 9280 in middle result but sum of 11136 and 9280 ("103936"). Implementation on C looks like this:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include "algutils.h"

#define ASCII2I(CH) ((CH)?((CH)-'0'):0)
#define I2ASCII(I) ((I)+'0')

char* mul(const char *s1, const char *s2, char *res, int reslen)
{
    int r, // index in result string
        i, // index in s1
        _i, // offset (to left) of next middle number
        j, // index in s2
        len1 = strlen(s1), len2 = strlen(s2);
    int d1, // digit in s1
        d2, // digit in s2
        ts, // temporary summ
        c, // curry in multiplication
        cs; // curry in summation
    memset(res, 0, reslen);
    for (i=len2-1, _i=0, cs=0; i>=0; i--, _i++) {
        d2 = ASCII2I(s2[i]);
        r = reslen - 2 - _i;
        for (j=len1-1, c=0; j>=0; j--) {
            d1 = ASCII2I(s1[j]);
            d1 = d1*d2 + c;
            ts = ASCII2I(res[r]) + (d1%10) + cs;
            res[r] = I2ASCII(ts%10);
            cs = ts/10;
            c = d1/10;
            r--;
        }
        do {
            // do rest of sum (high digits)
            ts = ASCII2I(res[r]) + (c%10) + cs;
            if (!ts) break;
            res[r] = I2ASCII(ts%10);
            cs = ts/10;
            r--;
        } while (cs);
    }
    return (res + r + 1);
}

///////////////////////////////////////////////
//              Main: tests
///////////////////////////////////////////////

#define RND(N) (srand(time(NULL)), rand()%(N))

int main(void)
{
    char _res[32], *res;
    char s1[32] = {0};
    char s2[32] = {0};
    int n1, n2, rn, resn, chk;

    //printf("%s\n", mul("1856", "1856", _res, sizeofarray(_res)));
    //return 0;

    for (int i=0; i<100; i++) {
        sprintf(s1, "%d", RND(9999999));
        sprintf(s2, "%d", 1+RND(9999999));
        printf("STR: %s * %s = %s\n", s1, s2, (res=mul(s1, s2, _res, sizeofarray(_res))));
        n1 = atoi(s1); n2 = atoi(s2); rn = n1*n2; resn = atoi(res);
        printf("CTR: %d * %d = %d\n", n1, n2, rn);
        printf("verify <%d>: %d (%d ? %d)\n", i, (chk=(resn==rn)), rn, resn);
        if (!chk) { printf("FAILED!\n"); return (1); }
        else printf("\n\n");
    }
    printf("All tests are passed\n");
    return (0);
}

Tests are cycle of multiplication of random numbers. If test is failed, cycle is break.