[Rd] Re: R-devel Digest, Vol 15, Issue 26
Josef Leydold
leydold at statistik.wu-wien.ac.at
Fri May 28 13:37:16 CEST 2004
Hi,
I am writing a package that provides an interface to other uniform
random number generators. It uses RNGkind(kind="user-supplied") and a
function double *user_unif_rand() as described in the help page
"Random.user". When looking into CRAN I found three other packages
that provide such a function in the DLL that they contribute:
rlecuyer, rsprng, and SuppDists (where it is currently disabled).
When loading two of thes then the functionality of at least one these
packages is affected. Moreover, a user cannot add his own random
number generator. Probably this is not a problem at all and will not
become a problem. Nevertheless, ...
Is it possible and desired to add the optinional argument PACKAGE to
RNGkind similarly to the .Call() function i.e. to have
RNGkind(kind="user-supplied", PACKAGE="mypackage") ?
When looking at the sources I think it should be straight forward
by extending
typedef struct {
RNGtype kind;
N01type Nkind;
char *name; /* print name */
int n_seed; /* length of seed vector */
Int32 *i_seed;
char *package; /* <-- added */
} RNGTAB;
and changing the corresponding calls in src/main/RNG.c.
If the anser is yes, maybe I could try to provide a patch for RNG.c.
The only thing I do not fully understand are the entries in
./src/main/names.c (but I have not studied the syntax of this file in
detail).
Josef
--
-----------------------------------------------------------------------------
Josef Leydold | University of Economics and Business Administration
| Department for Applied Statistics and Data Processing
-----------------------------------------------------------------------------
Augasse 2-6 | Tel. *43/1/31336-4695
A-1090 Vienna | FAX *43/1/31336-738
European Union | email Josef.Leydold at statistik.wu-wien.ac.at
-----------------------------------------------------------------------------
Alles Unglueck kam daher, dass die Denkenden nicht mehr handeln konnten,
und die Handelnden keine Zeit mehr fanden zu denken. (Marlen Haushofer)
More information about the R-devel
mailing list