On Thu, 27 Sep 2018, Sven Schreiber wrote:
 On Wed, 26 Sep 2018, Sven Schreiber wrote:
>> 
>>> OK, sorry for the delay. I'm attaching an attempt
 Minor correction: The included helper function matof (as in: matrix variant 
 of atof) had a little redundancy, because atof itself already handles empty 
 strings. Here's the update:
 <hansl>
 function matrix matof(strings in, int skip[0::0])
     # Returns column vector with the numerical values in the string array.
     # 'skip' refers to how many leading array elements should be discarded.
     N = nelem(in)
     matrix m = ones(N - skip,1) * $huge # initialize to nonsense
     loop i = 1+skip .. N --quiet
         m[i - skip] = atof(in[i])
     endloop
     return m
 end function
 </hansl> 
I was wondering, yesterday, when looking at your code: why initialise m to 
$huge? wouldn't
matrix m = mshape(NA, N - skip, 1)
do?
And another thing: it may make sense to expand atof() so as to return 
automatically a column vector if the argument is a string array. Or am I 
missing something?
-------------------------------------------------------
   Riccardo (Jack) Lucchetti
   Dipartimento di Scienze Economiche e Sociali (DiSES)
   Università Politecnica delle Marche
   (formerly known as Università di Ancona)
   r.lucchetti(a)univpm.it
   
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------