On 02/07/2010 03:58 PM, [email protected] wrote:
>
> Yes. The new api is our current focus.
>
> You can find many real examples as part of our unit tests (in the "unitTests"
> installed directory of the package).
>
> Some other examples in our blogs
>
> A draft article in the "papers" sub dir
ginal Message-
| From: Leo Alekseyev
| Date: Sun, 7 Feb 2010 08:11:48
| To: Romain Francois
| Cc:
| Subject: Re: [Rcpp-devel] Handling NAs in RcppMatrix
|
| D'oh, I should have guessed it was an integer/double issue -- I
| actually tried to check by testing something like c(1.0:2.2,NA), but
Original Message-
From: Leo Alekseyev
Date: Sun, 7 Feb 2010 08:11:48
To: Romain Francois
Cc:
Subject: Re: [Rcpp-devel] Handling NAs in RcppMatrix
D'oh, I should have guessed it was an integer/double issue -- I
actually tried to check by testing something like c(1.0:2.2,NA), but
this of
D'oh, I should have guessed it was an integer/double issue -- I
actually tried to check by testing something like c(1.0:2.2,NA), but
this of course still produced an integer.
So as a general question -- would you recommend using the new API
(Rcpp::CharacterVector, Rcpp::NumericVector) over RcppMat
I've logged this as a bug in Rcpp's bug tracker :
https://r-forge.r-project.org/tracker/?atid=637&group_id=155&func=browse
Romain
On 02/07/2010 10:15 AM, Romain Francois wrote:
> Hello,
>
> I think the problem is that c(1:2, NA) produces an __integer__ vector,
> and RcppVector does not take care
On 02/06/2010 11:56 PM, Leo Alekseyev wrote:
> Howdy folks,
[...]
>
> On a (probably) unrelated note, I get the following error when
> compiling my C++ code:
> Error in RcppCxx0xFlags() : could not find function "capture.output"
> Calls: -> cat -> RcppCxx0xFlags
I think I've fixed this in svn b
Hello,
I think the problem is that c(1:2, NA) produces an __integer__ vector,
and RcppVector does not take care of coercion properly.
> typeof( c(1:2,NA) )
[1] "integer"
> typeof( c(1,2,NA) )
[1] "double"
The "garbage value" you get I suppose is -2^31, which is how R
represents NA for intege
The behavior I'm seeing is a bit bizarre. The NaN conversion seems to
fail for RcppMatrix and RcppVector, but not for NumericVector. The
most interesting bit is this: if I run the test code on e.g.
c(NA,1,2), RcppVector works (NA is printed as nan), whereas if I use
c(NA,1:2), instead of nan it p
On 6 February 2010 at 22:08, Leo Alekseyev wrote:
| Thanks for the response. Your new unit test runs, but I still see the
| problem (running 0.7.4 installed from CRAN). Consider this version of
| the test -- the only difference is that it uses a 2x2 matrix and
| checks for NAs:
|
| test.RcppMat
Thanks for the response. Your new unit test runs, but I still see the
problem (running 0.7.4 installed from CRAN). Consider this version of
the test -- the only difference is that it uses a 2x2 matrix and
checks for NAs:
test.RcppMatrix.double.na <- function() {
src <- 'RcppMatrix m(x);
Hi Leo,
On 6 February 2010 at 17:56, Leo Alekseyev wrote:
| Howdy folks,
| I started playing around with Rcpp recently, and so far haven't been
| able to figure out how to detect NAs in a matrix that I pass to the
| C++ code.
| My test code looks something like this:
| RcppExport SEXP rcpp_test(S
Howdy folks,
I started playing around with Rcpp recently, and so far haven't been
able to figure out how to detect NAs in a matrix that I pass to the
C++ code.
My test code looks something like this:
RcppExport SEXP rcpp_test(SEXP N1, SEXP V1, SEXP M1, SEXP parms) {
...
RcppMatrix m1(M1);//
12 matches
Mail list logo
