[Scilab-users] eomday

[Stefan Du Rietz]

Hi,
there is a bug in eomday.

+ If there is a leap year between two ordinary years, e.g.

-->Y = [2011 2012 2013]'
  Y  =
     2011.
     2012.
     2013.
-->M = [10 11 3]'
  M  =
     10.
     11.
     3.
-->eomday(Y, M)
  !--error 15
Submatrix incorrectly defined.
at line      31 of function eomday called by :
eomday(Y, M)

+ Lines 28-33 of the function code, with my "pause" added at 29:

28  [nr,nc] = size(M);
29  pause
30  E(  isLeapYear(Y) ) = leap_year(   M( isLeapYear(Y)) );
31  E( ~isLeapYear(Y) ) = common_year( M(~isLeapYear(Y)) );
32	
33  E = matrix(E,nr,nc);

-->eomday(Y, M)

-1->whereami
whereami called under pause
pause    called at line 29 of macro eomday

-1->E(  isLeapYear(Y) ) = leap_year(   M( isLeapYear(Y)) )
  E  =
     0.
     30.
-1->E( ~isLeapYear(Y) ) = common_year( M(~isLeapYear(Y)) );
E( ~isLeapYear(Y) ) = common_year( M(~isLeapYear(Y)) )
                                                        !--error 15
Submatrix incorrectly defined.

-1->size( ~isLeapYear(Y) )
  ans  =
     4.    1.

+ Because E is not pre-dimensioned, the index is longer than the vector.

+ Make E a vector with an equal number of elements as M:

-1->E = zeros(1,nr*nc);

+ Now the next lines work:

-1->E(  isLeapYear(Y) ) = leap_year(   M( isLeapYear(Y)) );
  E  =
     0.    30.    0.    0.
-1->E( ~isLeapYear(Y) ) = common_year( M(~isLeapYear(Y)) )
  E  =
     31.    30.    31.    31.
-1->E = matrix(E,nr,nc)
  E  =
     31.
     30.
     31.
     31.

+ So, a new line 29 is needed:

29  E = zeros(1,nr*nc);

Regards
Stefan