mfile2sci: the only-double option should be %F by default
Reported by Samuel GOUGEON (@sgougeon)
BUG DESCRIPTION:
----------------
The fact that the only-double input argument is %T is (too) restrictive and has unknown consequences.
It is presently described as
"Boolean flag, if %T mfile2sci considers that numerical function have been used only with
numerical data (no Scilab overloading function is needed). Default value: %T"
This demands to be aware of all possible input types in function AND operators.
I was trying to convert simply
s = 'ab' + 'cd'
and it gave me
s = "ab" + "cd"
while i was expecting
s = asciimat("ab") + asciimat("cd")
It was puzzling. Only debugging this issue has led me to the only_double option
in convert2double(), and then to the mfile2sci() only-double one.
The default configuration of the converter should be less restrictive,
even if it then runs more slowly (if this is a consequence, it should be documented).
By the way, the impact of this mfile2sci option on the addition or substraction should
be documented in the mtlb_a and mtlb_s pages.