[Rd] Using unicode with sprintf or paste in Windows (PR#11515)

[ripley at stats.ox.ac.uk]

This is not a bug -- it is the expected (and documented) behaviour.

On Mon, 26 May 2008, mwtoews at sfu.ca wrote:

> Full_Name: Michael Toews
> Version: 2.7.0
> OS: Windows XP SP2
> Submission from: (NULL) (24.80.163.230)
>
>
> Using Unicode characters in Windows works with static strings (as of R 2.7.0),
> however fails when used with sprintf() or paste(). For example, on R 2.7.0 for
> Windows XP (SP2):
>
> Static string (works, showing Greek Delta character):
>> print("\u0394Q = 2.2 L/s")
> [1] "ΔQ = 2.2 L/s"
>
> However, when I use paste() or sprintf(), the Unicodeness is lost:
>> paste("\u0394Q =",round(1.2345,3),"L/s")
> [1] "<U+0394>Q = 1.234 L/s"
>> sprintf("\u0394Q = %.3f L/s",1.2345)
> [1] "<U+0394>Q = 1.234 L/s"
>
>
> I have also tested this with R 2.7.0 for Linux, which yields the following:
>> print("\u0394Q = 2.2 L/s")
> [1] "&#916;Q = 2.2 L/s"
>> paste("\u0394Q =",round(1.2345,3),"L/s")
> [1] "&#916;Q = 1.234 L/s"
>> sprintf("\u0394Q = %.3f L/s",1.2345)
> [1] "&#916;Q = 1.234 L/s"

It depends on the locale ... you cannot expect to be able to manipulate 
Unicode strings in a non-Unicode locale (although you can pass them around 
as data, unchanged).

> (Note: this was posted on R-help with no replies:
> https://stat.ethz.ch/pipermail/r-help/2008-May/162815.html )

Well, the posting guide asks you to do your homework before posting.
In this case, the 'R Internals' manual and the source code.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595