An Alternative Conflict Resolution Strategy

R's default conflict management system gives the most recently loaded package precedence. This can make it hard to detect conflicts, particularly when they arise because a package update creates ambiguity that did not previously exist. 'conflicted' takes a different approach, making every conflict an error and forcing you to choose which function to use.

Travis buildstatus Coveragestatus CRANstatus

The goal of conflicted is to provide an alternative conflict resolution strategy. R’s default conflict resolution system gives precedence to the most recently loaded package. This can make it hard to detect conflicts, particularly when introduced by an update to an existing package. conflicted takes a different approach, making every conflict an error and forcing you to choose which function to use.

Thanks to @krlmlr for this neat idea! This code was previously part of the experimental strict package, but I decided improved conflict resolution is useful by itself and worth its own package.


# install.packages("devtools")


To use conflicted, all you need to do is load it:

filter(mtcars, cyl == 8)
#> Error: [conflicted] `filter` found in 2 packages.
#> Either pick the one you want with `::` 
#> * dplyr::filter
#> * stats::filter
#> Or declare a preference with `conflicted_prefer()`
#> * conflict_prefer("filter", "dplyr")
#> * conflict_prefer("filter", "stats")

As suggested, you can either namespace individual calls:

dplyr::filter(mtcars, am & cyl == 8)
#>    mpg cyl disp  hp drat   wt qsec vs am gear carb
#> 1 15.8   8  351 264 4.22 3.17 14.5  0  1    5    4
#> 2 15.0   8  301 335 3.54 3.57 14.6  0  1    5    8

Or declare a session-wide preference:

conflict_prefer("filter", "dplyr")
#> [conflicted] Will prefer dplyr::filter over any other package
filter(mtcars, am & cyl == 8)
#>    mpg cyl disp  hp drat   wt qsec vs am gear carb
#> 1 15.8   8  351 264 4.22 3.17 14.5  0  1    5    4
#> 2 15.0   8  301 335 3.54 3.57 14.6  0  1    5    8

I recommend declaring preferences directly underneath the corresponding library call:

conflict_prefer("filter", "dplyr")

You can ask conflicted to report any conflicts in the current session:

#> 6 conflicts:
#> * `filter`   : [dplyr]
#> * `intersect`: [dplyr]
#> * `lag`      : dplyr, stats
#> * `setdiff`  : [dplyr]
#> * `setequal` : [dplyr]
#> * `union`    : [dplyr]

Functions surrounded by [] have been chosen using one of the built-in rules. Here filter() has been selected because of the preference declared above; the set operations have been selected because they follow the superset principle and extend the API of the base equivalents.

How it works

Loading conflicted creates a new “conflicted” environment that is attached just after the global environment. This environment contains an active binding for any object that is exported by multiple packages; the active binding will throw an error message describing how to disambiguate the name. The conflicted environment also contains bindings for library() and require() that suppress conflict reporting and update the conflicted environment with any new conflicts.

Alternative approaches

It is worth comparing conflicted to modules and import. Both packages provide strict alternatives to library(), giving much finer control over what functions are added to the search path.

# modules expects you to namespace all package functions
dplyr <- modules::import_package('dplyr')
dplyr$filter(mtcars, cyl == 8)
# import expects you to explicit load functions
import::from(dplyr, select, arrange, dplyr_filter = filter)
dplyr_filter(mtcars, cyl == 8)

These require more upfront work than conflicted, in return for greater precision and control.

Code of conduct

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.


conflicted 1.0.1

  • Internal has_moved() function no longer fails when it encounters a call to .Deprecated() with no arguments (#29).

  • .conflicts environment is correctly removed and replaced each time a new package is loaded (#28).

conflicted 1.0.0

New functions

  • conflict_scout() reports all conflicts found with a set of packages.

  • conflict_prefer() allows you to declare a persistent preference (within a session) for one function over another (#4).

Improve conflict resolution

  • conflicts now generally expects packages that override functions in base packages to obey the "superset principle", i.e. that foo::bar(...) must return the same value of base::bar(...) whenever the input is not an error. In other words, if you override a base function you can only extend the API, not change or reduce it.

    There are two exceptions. If the arguments of the two functions are not compatible (i.e. the function in the package doesn't include all arguments of the base package), conflicts can tell it doesn't follow the superset principle. Additionally, dplyr::lag() fails to follow the superset principle, and is marked as a special case (#2).

  • Functions that have moved between packages (i.e. functions with a call to .Deprecated("pkg::foo")) as the first element of the function body) will never generate conflicts.

  • conflicted now listens for detach() events and removes conflicts that are removed by detaching a package (#5)

Minor improvements and bug fixes

  • Error messages for infix functions and non-syntactic function names are improved (#14)

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


1.0.4 by Hadley Wickham, 2 years ago

Report a bug at

Browse source code at

Authors: Hadley Wickham [aut, cre] , RStudio [cph]

Documentation:   PDF Manual  

GPL-3 license

Imports rlang, memoise

Suggests covr, crayon, dplyr, pkgdown, testthat

Imported by eat, rUM, tidymodels.

Suggested by DGEobj, DGEobj.utils, DOPE, boxr, here, vegawidget, vembedr, votesmart.

See at CRAN