watch: Instrument a Function for Variable Watching
In brodieG/watcher: Watch Function State

Description Usage Arguments Details Value Note See Also Examples

The input function will be modified such that the state of the function environment is captured after each top-level statement is evaluated. Top-level statements are, roughly speaking, distinct R expressions that are visible directly in the source of the R function.

1	watch(fun, vars = character())

`fun`	a function to watch
`vars`	character a vector of names of variables to record

For each top-level step in the evaluation of a watched function a list element is added to the "watch.data" attribute of the result. The list will contain a copy of all the variables in the function environment right after that step is evaluated, or of the subset of them specified by the vars parameter. Additionally, the list will contain a "line" attribute that maps to the start line of the top-level expression that was last evaluated. This line number maps to the deparsed function attached to the result as the "watch.code" attribute. The line numbers may or may not match to other deparsings of the function, so if you intend on using the line numbers be sure to do so in relation to the "watch.code" version of the function.

Instrumented function semantics should be the same as the un-instrumented version, except for the attributes attached to the return value, and for the use of the .res temporary variable. If a function also uses the .res symbol the instrumentation will interfere with the original semantics.

an instrumented version of fun. When this instrumented function is run it will add attributes "watch.data" and "watch.code" to the result. See the description for details about return data format of the instrumented function.

if you are watching a function from a package you might want to install the package with Sys.setenv('R_KEEP_PKG_SOURCE'='yes') so that the line numbers match the original source, although keep in mind this will make the installation larger (see ?options and search for keep.source.pkgs).

simplify_data(), vignette('watcher', package='watcher').

insert_sort2 <- watch(insert_sort, c('x', 'i', 'j'))
res <- insert_sort2(runif(10))
dat <- simplify_data(attr(res, 'watch.data'))
code <- attr(res, 'watch.code')