Vtgyjfy

I would like to limit floating point numbers inside a label:

beginfig(1)

   for i=1 upto 5:

      label.bot(textext("$" & (decimal (-0.8*i)) & "$"), (-0.8*i, -0.2));

endfig;

end;

Currently it shows 5 numbers after comma. I need 2.

You can define a MetaPost macro to select the right substring:

vardef tdecimal(expr num, d) :=

  save numS; string numS; numS = decimal(num);

  save j; j := length numS;

  for k = 0 upto length numS - 1:

    if "." == substring (k,k+1) of numS: j:=k; fi

  endfor

  for k = j + 1 upto j + d:

    if "0" <> substring (k,k+1) of numS: j:=k+1; fi

  endfor

  substring (0,j) of numS

enddef;

Then you can use e.g. tdecimal(1.234567, 2) to get 1.23 or tdecimal(1.234567, 3) to get 1.234.

An easier alternative would be to separately get the string representation of the integral and the decimal part, but this leads to some rounding problems in MetaPost.
Anyway, if you want to try it, the code would be

vardef twodecimal(expr num) :=

  if num < 0:

    "-" & decimal(floor(-num)) & "." & substring (1,3) of decimal(floor((-num-floor(-num)+1)*100))

  else:

    decimal(floor(num)) & "." & substring (1,3) of decimal(floor((num-floor(num)+1)*100))

  fi

enddef;

Add usepackage{xintfrac} to preamble (prologue? I don't remember metapost terminology) and use

label.bot(textext("$xintRound{2}{" & (decimal (-0.8*i)) & "}$"), (-0.8*i, -0.2))

Well I did your job and I opened up MetaPost manual. I don't know what is your "textext", which must be provided by some additional macros.

input TEX;



TEXPRE("input xintfrac.styrelax");



beginfig(1);

pickup pencircle scaled 1pt;

draw (0,0);

numeric i;

i := 1.236;

label.bot(TEX("$xintTrunc{2}{" & (decimal (-0.8*i)) & "}$"), (-0.8*i, -0.2))

endfig;



end

When acted upon by mpost produces a correct EPS file. I used xintTrunc as you want truncation. That could be done with no package at all (truncation is easier than rounding.)

I searched MetaPost manual for all occurences of "decimal" and did not see how to do the truncation before hand, but that must be possible because among the references I see "decNumber ANSI C implementation of general decimal arithmetic".