Re: ran into ambiguous code (bug?)

Hello,

Inline.

Às 06:53 de 28/07/2026, Martin Maechler escreveu:
> Dear Ben,
>
> As this may be instructive for some in the R-help audience,
> I'm replying here in public (and Bcc to you):
>
>>>>>> Ben H....  on Mon, 27 Jul 2026 16:50:20 -0700 writes:
>      > Hi,
>      > I would like to submit a bug report or at least get some help.  I've used R
>      > for about 10 years and not reported a bug before.  Today I ran into
>      > ambiguous R code, the code is reading non-existent columns.  A reproducible
>      > example of the bug along with the R version report is below.  'Age' does
>      > not exist among the column names of the data frame (df), but 'Age2' does.
>      > When you run df$Age, it is equivalent to df$Age2. I have reproduced this
>      > issue after removing all packages.
>
>      > 1. What should I do to use R without this issue?
>      > 2. Can/should I submit this?
>
>      > R version 4.5.2 (2025-10-31) -- "[Not] Part in a Rumble"
>      > Copyright (C) 2025 The R Foundation for Statistical Computing
>      > Platform: aarch64-apple-darwin20
>
>      ......
>
>      >> df <- data.frame(people = c('Susan', 'Mike', 'Gary', 'Wayne'),
>      >                   Age2 = c(2,24,35,1))
>      >>
>      >> df$Age2
>      > [1]  2 24 35  1
>      >> df$Age
>      > [1]  2 24 35  1
>      >>
>      >> "Age2" %in% colnames(df)
>      > [1] TRUE
>      >> "Age" %in% colnames(df)
>      > [1] FALSE
>      >>
>
>      > --
>      > Ben H....., PhD
>
> Dear Ben,
>
> You are reporting a problem you have with R, but not at all bug:
>
> This is how R has worked forever (and S & S-plus before that).
>
> Note:
>   - column names can be abbreviated when using `$` (hence you can
>     use `$Age` equivalently to `$Age2`
>   - Since this maybe confusing to some, you can make it to a warning  via
>
>           options(warnPartialMatchDollar = TRUE)
>
> Historical reason:
> In earlier times, before people had IDEs such as Rstudio etc,
> VScode, or Emacs(ESS) to do data analysis in R, writing long
> variable names was cumbersome and hence allowed to be abbreviated.
>
> In your example :
>
>> options(warnPartialMatchDollar = TRUE)
>> df$Age
> [1]  2 24 35  1
> Warning message:
> In df$Age : partial match of 'Age' to 'Age2'
>
>
> Note that I recommend using   df[,"Age2"]   (which works in your
> example) whereas  df[,"Age"] gives an error :
>
>> df[,"Age"]
> Error in `[.data.frame`(df, , "Age") : undefined columns selected


Or df[["Age"]], with a different result.


df <- data.frame(people = c('Susan', 'Mike', 'Gary', 'Wayne'),
                  Age2 = c(2,24,35,1))

df[, "Age"]   # error
#> Error in `[.data.frame`:
#> ! undefined columns selected
df[, "Age2"]
#> [1]  2 24 35  1

df[["Age"]]   # no error
#> NULL
df[["Age2"]]
#> [1]  2 24 35  1

fun <- function(data, col_name) {
   x <- data[[col_name]]    # if a warning is OK
   # x <- data[, col_name]  # if you want to get an error
   if(is.null(x)) {
     msg <- sprintf("column %s not found, returning NULL", sQuote(col_name))
     warning(msg)
   }
   x
}

fun(df, "Age")
#> Warning in fun(df, "Age"): column 'Age' not found, returning NULL
#> NULL
fun(df, "Age2")
#> [1]  2 24 35  1

The rule of thumb is to use `$` in interactive sessions but `[` or `[[` 
when programming, like in the function above.


These two StackOverflow posts may be relevant [1] and [2].


[1] 
https://stackoverflow.com/questions/18222286/dynamically-select-data-frame-columns-using-and-a-character-value

[2] 
https://stackoverflow.com/questions/1169456/the-difference-between-bracket-and-double-bracket-for-accessing-the-el


Hope this helps,

Rui Barradas



>
>
> We have looked into this (and related  "partial matching"), and
> found that it would "break" too much old R code, if the warning
> was turned on unconditionally.
>
> You may be interested in reading  help(options)  and search for
> other 'warnPartialMatch*' ..
> e.g., only at
>    https://cran.r-project.org/doc/manuals/r-release/packages/base/refman/base.html#options
>
> Best regards,
> Martin
>
>
> --
> Martin Maechler
> R Core team  and  ETH Zurich
>
> ______________________________________________
> [email protected] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide https://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.