Update Attributes.R (closes #1017)

[tjmckinley]

When passing multiple "depends" arguments to "cppFunction" e.g.

cppFunction(code, depends = c("package1", "package2"))

this was producing code with scaffolding:

// [[Rcpp:depends(package1)]]
// [[Rcpp:depends(package2)]]
#include <package1.h>
#include <Rcpp.h>

rather than

// [[Rcpp:depends(package1, package2)]]
#include <package1.h>
#include <package2.h>
#include <Rcpp.h>

[coatless]

Good catch! LGTM.

[tjmckinley]

No worries. Thanks to all for producing such a useful package!

[coatless]

@tjmckinley while you're at it... Mind fixing the code for plugins as well? c.f.

Rcpp/R/Attributes.R

Lines 258 to 259 in 975361f

	plugins <- paste(plugins, sep=", ") # #nocov start
	pluginsAttrib <- paste("// [[Rcpp::plugins(", plugins, ")]]", sep="")

[eddelbuettel]

@tjmckinley Thank you. Please also consider reading Contributing before sending PRs.

What is the motivation for the PR? What it fixes is not a defect but an aesthetic preference. Which would change something used as it currently is and was by lots of people. We generally don't do that without good reason.

[tjmckinley]

My apologies @eddelbuettel. I will log an issue now with a reprex.

[tjmckinley]

Hi @eddelbuettel. I've logged an issue (#1017) with a reprex. It's not purely aesthetic in this case since the code won't compile for the example since the header file for RcppArmadillo is not found. For example, running a function of the form

cppFunction(code, depends = c("package1", "package2"))

produces code with scaffolding:

// [[Rcpp:depends(package1)]]
// [[Rcpp:depends(package2)]]
#include <package1.h>
#include <Rcpp.h>

with no #include <package2.h>. This is caused simply by the cppFunction() code not parsing the depends argument correctly.

Many thanks,

TJ

[tjmckinley]

By the way, many thanks @eddelbuettel, @coatless for producing this fantastic package. I can't do without it! Apologies also if I take a while to reply. I'm in Japan and hence the time difference causes a lag in response time.

[tjmckinley]

@tjmckinley while you're at it... Mind fixing the code for plugins as well? c.f.

Rcpp/R/Attributes.R

Lines 258 to 259 in 975361f

	plugins <- paste(plugins, sep=", ") # #nocov start
	pluginsAttrib <- paste("// [[Rcpp::plugins(", plugins, ")]]", sep="")

Yes @coatless. Happy to do this if you and @eddelbuettel think this is an issue.

[jjallaire]

LGTM as well.

[eddelbuettel]

Looks good to me too after a closer look; merging.