R devel - Dec 2020 - translation domain is wrong in stopifnot()?

I don't know if this would be considered a bug in either stopifnot()
or (n)gettext(), or not a bug at all, but sometimes the translation
domain is not set properly for stopifnot() messages, so they won't be
translated. E.g.

Sys.setenv(LANGUAGE = "de")

# This is good
stopifnot(FALSE)
#> Fehler: FALSE ist nicht TRUE

f <- function() stopifnot(FALSE)
# This is not
f()
#> Fehler in f() : FALSE is not TRUE

Gabor

>>>>> G?bor Cs?rdi 
>>>>>     on Tue, 1 Dec 2020 23:48:37 +0000 writes:
    > I don't know if this would be considered a bug in either
stopifnot()
    > or (n)gettext(), or not a bug at all, but sometimes the translation
    > domain is not set properly for stopifnot() messages, so they won't
be
    > translated. E.g.

    > Sys.setenv(LANGUAGE = "de")

    > # This is good
    > stopifnot(FALSE)
    > #> Fehler: FALSE ist nicht TRUE

    > f <- function() stopifnot(FALSE)
    > # This is not
    > f()
    > #> Fehler in f() : FALSE is not TRUE

    > Gabor

or even just directly :

  > (function() stopifnot(1))()
  Fehler in (function() stopifnot(1))() : 1 is not TRUE


I agree there is a bug.

The problem is that the default for the optional 'domain'
argument does not "work" correctly in the 2nd case.

I've checked that this can be worked around if I explicitly add

  , domain = "R-base"

to the ngettext() call   [ and also the gettext() call above ] .
Then I do get

  > (function() stopifnot(1))()
  Fehler in (function() stopifnot(1))() : 1 ist nicht TRUE
  > 

For now I'd claim the bug is in the underlying C code of
gettext() , ngettext() ...

It would we good to report this in R's bugzilla, please,
see https://www.r-project.org/bugs.html

Martin