print() support for variable arguments

[jeremydouglass]

#176 adds support for redirecting print() to console -- this works well for one argument.

In Processing(Java), print supports varargs -- e.g. void print(Object... arg){}, like this:

print("a", 1, "b", mouseX);

a 1 b 75

You can see it in the docs here:

• https://processing.org/reference/print_.html

Two questions:

1. How hard would it be to support this in Processing.R? It isn't high priority, but it is common when logging to write for example print(mouseX, mouseY);
2. If it could be supported, what is the most natural way to write varargs variable length argument lists in R? Would it just be print("a", 1, "b", mouseX)?

[gaocegege]

I am not familiar with the print system in R, so I need to try it.

[jeremydouglass]

This looks complex.

The R native print method:

print.default() is the default method for the S3 generic print() function. That means that to call print.default(), you just call print() on an object that doesn't have its own print() method. https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/print.default

... is then mapped by renjin like this:

https://github.com/bedatadriven/renjin/blob/master/core/src/main/java/org/renjin/primitives/Print.java

So the question is, should Processing.R mode be giving full access to the Processing(Java) print() method -- which uses varargs -- or should it be using the native R print() -- which accepts a single object, but that object can be a vector or a complex data frame?

[gaocegege]

I think P.R should support these two ways, since The users comes from both Processing and R community.

[jeremydouglass]

Interesting. Supporting both methods would either require:

1. multiple signatures for print that could detect a data primitive vs. vector / data frame vs. varargs, or
2. renaming one of print methods -- for example "print / pprint" for Processing-print, or "print / rprint" for R-print.

Which approach do you think makes more sense? Or is there another approach?

[jeremydouglass]

Thinking about this further, I think that "print" in R mode should be left in an R-native style -- it takes a single object, which may be a vector.

To make this easier for learners, I could:

1. generate a helpful warning message about how to use vector-style if print() is given the wrong number of arguments. This would particularly help people who are cut-pasting Java mode or python mode sketches into R-mode and trying to adapt them one error at a time.

2. update the documentation to explain usage at https://processing-r.github.io/reference/print.html