metaprogramming-rlang

To pass column names as arguments in custom dplyr functions, use tidy evaluation to embrace them with {{ }}. This forwards function arguments safely into data-masking verbs like summarise() or mutate() without hard-coding names.

Tidy evaluation is a framework in R that enables programmatic tidyverse APIs by bridging data-masking and selection patterns. You need it when building reusable functions that accept dynamic column names or inject expressions at runtime.

Injection fails when lists are not properly spliced. To inject a list of columns into mutate, use the !!! operator for splicing, ensuring dynamic dots and expressions are safely evaluated within the data-masking environment.

In rlang data-masking, the .data pronoun explicitly retrieves objects from the data frame, while .env retrieves objects from the execution environment. Using both ensures safe, name-stable, and robust programmatic code.

Yes, you can use dynamic dots to wrap generic tidyverse functions for arbitrary column selections. By passing ... and leveraging injection operators, your function can operate on any columns provided at runtime.

Limitations of rlang tidy eval include a steep learning curve for syntax like embrace, injection, and pronouns. If functions lack proper .data or .env pronouns, code may become name-unstable or fail to resolve masking ambiguities safely.