Using errorize

Ian K. Kyle



The errorizer package allows one to create drop-in replacements for existing R functions, but with added logging and handling functionality when encountering errors or warnings. When working interactively, it is fairly easy to debug most errors because one can access any and all objects entering a function call. When using R CMD BATCH, Rscript, or when running code in parallel, one often does not have the ability to immediately resurrect the objects that have entered a function call for debugging purposes.

The errorize function accepts an existing R function as its first argument and returns a R function with the exact same arguments and functionality. However, if an error or warning occurs when running that “errorized” R function, it will save a .Rds file to the current working directory with the relevant objects and information required to immediately recreate the error.


“Errorizing” a function

Use errorize() to create an “errorized” function:

mean2 <- errorize(mean)

The function mean2 now works as mean does:

## [1] 5.5

But it will now write to a .Rds file if it issues a warning:

Warning message:
In value[[3L]](cond) :
  Wrote to ./mean_warning_1478361326.Rds on catching "simpleWarning in mean.default(x = c("a", "b", "c", "d", "e", "f", "g", "h", "i", : argument is not numeric or logical: returning NA

If you want to create a drop-in replacement for an existing function, you should specify the function’s namespace in the call to errorize to avoid conflicts.

lm <- errorize(stats::lm)
lm(speed ~ dist, data = cars) # this works
lm(speed ~ missing_variable, data = cars) # this errors out and saves the data
Error in value[[3L]](cond) : 
  Wrote to ./lm_error_1478361734.Rds on catching "Error in eval(expr, envir, enclos): object 'missing_variable' not found

To read that data back in and debug, use readRDS().

er <- readRDS('./lm_error_1478361734.Rds')

Recreate the error using$fxn, er$arglst)

You can inspect the objects that entered the function call, they are all separate elements of a named list that is the arglist slot of the saved object.

Additional arguments

An “errorized” function will by default halt progress if the function issues an error, but you can set the errorize argument stopOnError to FALSE. In this case, the function will issue a warning instead of an error, and will return the value specified in the argument onErrorReturn (default is NULL).


This is useful for “errorizing” one or several functions in a script, but if you want this type of behavior for all function calls, you might want to use options(error = quote(dump.frames("dump", TRUE))). See ?debugger (from the utils package).

This method won’t work if an object passed to the errorizer function does not exist since errorize evaluates all arguments in the parent.frame before passing them to the relevant function.

This method will still work if objects from the global environment are needed for the function to work but aren’t passed explicitly to the errorized function, but those objects will not be saved out to the .Rds file in the case of an error or warning.