Matt Weinberg <[email protected]> writes,
> This seems to work well, except for the proposed subroutine callf_setup.
> The suggested code for this routine was:
>
> pointer vector callf_setup(pointer(function) scalar f, real scalar n,
> a1, a2, a3, a4, a5)
> {
> pointer vector p
> real scalar mycopy
>
> p = J(1, 5+2, .)
>
> p[1] = f
> mycopy = n ; p[2] = &mycopy
> p[3] = &a1 ; p[4] = &a2 ; p[5] = &a3 ; p[6] = &a4 ; p[7] = &a5
>
> return(p)
> }
>
>
> but this will generate error 3253, "non-real found where real expected".
Change
p = J(1, 5+2, .)
to
p = J(1, 5+2, NULL)
I admit the error message is misleading, but what happened was that Mata
did not notice the problem when it executed -p = J(1, 5+2, .)-.
Instead, it made p a real vector. Then later, when Mata tried to
assign -p[1] = f-, it did notice the problem. By this point, p was real,
and so the error-message-generating routine decided to say that a "non-real
was found where real expected" when what in fact happened was that a real
was found earlier where a nonreal was expected.
Anyway, the point of
p = J(1, 5+2, NULL)
is to make p a 1x7 vector filled in with missing. NULL is a pointer missing.
-- Bill
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
