r/man/register_binding.Rd

% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/dplyr-funcs.R
\name{register_binding}
\alias{register_binding}
\title{Register compute bindings}
\usage{
register_binding(fun_name, fun, notes = character(0))
}
\arguments{
\item{fun_name}{A string containing a function name in the form \code{"function"} or
\code{"package::function"}.}

\item{fun}{A function, or \code{NULL} to un-register a previous function.
This function must accept \code{Expression} objects as arguments and return
\code{Expression} objects instead of regular R objects.}

\item{notes}{string for the docs: note any limitations or differences in
behavior between the Arrow version and the R function.}
}
\value{
The previously registered binding or \code{NULL} if no previously
registered function existed.
}
\description{
\code{register_binding()} is used to populate a list of functions that operate on
(and return)
Expressions. These are the basis for the \code{.data} mask inside dplyr methods.
}
\section{Writing bindings}{

\itemize{
\item \code{Expression$create()} will wrap any non-Expression inputs as Scalar
Expressions. If you want to try to coerce scalar inputs to match the type
of the Expression(s) in the arguments, call
\code{cast_scalars_to_common_type(args)} on the
args. For example, \code{Expression$create("add", args = list(int16_field, 1))}
would result in a \code{float64} type output because \code{1} is a \code{double} in R.
To prevent casting all of the data in \code{int16_field} to float and to
preserve it as int16, do
\code{Expression$create("add", args = cast_scalars_to_common_type(list(int16_field, 1)))}
\item Inside your function, you can call any other binding with \code{call_binding()}.
}
}

\keyword{internal}