>>>>> Randy Cragun
>>>>> on Thu, 30 May 2019 00:26:38 -0400 writes:
> I do not know if this is a bug or a case of improper
> documentation. The documentation for formatC() implies
> that the difference between the options format="f" and
> format="g" is that with "g", scientific format is
> sometimes used. There is another difference between them
> that is not mentioned in the
> documentation. drop0trailing=FALSE is ignored when format
> is set to "g" unless flag contains "#" (this is the
> documented behavior for format="fg"). For instance, the
> first line below return " 2.5", whereas the second returns
> the expected "2.50".
> formatC(2.50, format="g", digits=3, drop0trailing=F)
> formatC(2.50, format="g", digits=3, drop0trailing=F,
flag="#")
Well, you have a point that this behavior is not documented in
details (and I assume the text reference "Kernighan and Richie"
is less available for the typical R users than in 1995...)
However, formatC() has been unchanged like that for close to 20
years, so we will most probably not change the function's behavior.
Notice that drop0trailing=FALSE is really the default
(and format="g" is also the default for non-character / non-integer
numbers).
The design of formatC(*) for numbers has entailed to default to
format="g" which drops trailing zeros most of the time
[whereas the format = "f" does not, unless drop0trailing=TRUE is set.]
Lastly, note that 2.50 and 2.5 are exactly identical as R
numbers; so, your two examples above are identical to the much shorter
formatC(2.5, digits=3)
formatC(2.5, digits=3, flag="#")
If you want "extraneous" trailing zeros, the "f" format is
your
friend most of the time anyway:
> t(sapply(1:8, function(D) formatC(c(2.5,pi), format="f", digits=
D)))
[,1] [,2]
[1,] "2.5" "3.1"
[2,] "2.50" "3.14"
[3,] "2.500" "3.142"
[4,] "2.5000" "3.1416"
[5,] "2.50000" "3.14159"
[6,] "2.500000" "3.141593"
[7,] "2.5000000" "3.1415927"
[8,] "2.50000000" "3.14159265">
I will add more information to the formatC() help
page, notably not only mentioning but explaining most of the
'flag's that are available typically(*).
Thank you for raising the issue.
Martin Maechler
ETH Zurich and R Core Team
--
*) as formatC() interfaces to the OS C library, some of the
availability and meaning of 'flags' is platform dependent.
> ----------------------
> sessionInfo():
> R version 3.5.3 (2019-03-11) Platform:
> x86_64-w64-mingw32/x64 (64-bit) Running under: Windows > > 8
x64 (build 9200)
> Matrix products: default
> locale: [1] LC_COLLATE=English_United States.1252
> LC_CTYPE=English_United States.1252 [3]
> LC_MONETARY=English_United States.1252 LC_NUMERIC=C
> [5] LC_TIME=English_United States.1252
> attached base packages: [1] stats graphics grDevices utils
> datasets methods base
> loaded via a namespace (and not attached): [1]
> compiler_3.5.3 tools_3.5.3
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel