Piping code into cppFunction()?

[richgoldberg]

It seems that piping text containing a C++ function into cppFunction() doesn't lead to the function becoming part of the environment. (code is the first argument fo cppFunction() so I thought that it should work.)

Here's an example that didn't work for me:

'int fibonacci(const int x) {
    if (x == 0) return(0); 
    if (x == 1) return(1);
    return (fibonacci(x - 1)) + fibonacci(x - 2);
    }' %>% cppFunction()

In contrast, this worked perfectly for me:

cppFunction(
    'int fibonacci(const int x) {
        if (x == 0) return(0); 
        if (x == 1) return(1);
        return (fibonacci(x - 1)) + fibonacci(x - 2);
    }')

(I am using R version 3.4.0 on 64 bit Windows 7 with version 0.12.10 of Rcpp.)

[eddelbuettel]

Funny how using something the way it is actually documented works. News at 11::00pm.

[richgoldberg]

Dr. Eddelbuettel, I love the Rcpp package so don’t want to be offensive but is there a reason you wouldn’t want the Rcpp functions to be compatible with magrittr/dplyr? Your reply to my Issue seems like you dismiss the issue that some people may want to create their C++ code programmatically and would like to pipe the result into cppFunction(). I would understand if you said that the difficulty in making it compatible with the magritter pipe wasn’t worth the effort or if I was doing something wrong in how I was passing the arguments but your sarcastic response really seemed uncalled for. Yes, I can get the function to work which is all I personally need. However, your function doesn’t’ behave as expected in the case I identified so it seemed like a legitimate Issue to flag. Rich

[eddelbuettel]

You seem to imply that we owe you a function for piping into cppFunction(). We don't.

But this is Open Source. You can go crazy and write such a function, or even a package.

Or heck, go crazy and fork Rcpp itself into tidyRcpp or whatnot.

But your imagination went overboard when you convinced yourself that we ever suggested to use cppFunction() this way. We have not. We (and I say this as acting maintainer) are not all that likely to ever do that either. Allow me to repeat that a package working as documented is not in fact a bug.

[kevinushey]

More to the point -- if something works without magrittr's pipe, but does not work with magrittr's pipe, then that is surely a bug in magrittr, and the bug should be filed on their side.

[eddelbuettel]

And for closure, quoting @spacedman from a conversation in the StackOverflow chatroom:

its only because Rcpp has side effects that it fails :) You can still do fib = code %>% cppFunction()
or code %>% cppFunction(env=.GlobalEnv) will work for that user

Still all user error and runaway imagination...