[R] Sourcing files with Umlaut in path no longer works

[Ulrike Grömping]

Duncan and Peter, thank you very much!
Actually, I made a mistake with the path, due to my inexperience with 
Windows 7 (user instead of users). The R warning then confused me, 
because I was told that 'C:/user/grömping/documents/publicat/...' 
wasn't found.
In fact, once the path was correct, source worked without problems.
Thus, only the warning message did not properly handle the encoding.

Best regards,
Ulrike Grömping

Am 21.11.2012 15:59, schrieb Duncan Murdoch:
> On 21/11/2012 6:49 AM, Ulrike Grömping wrote:
>> Dear helpeRs,
>>
>> on my new machine - Windows 7 64Bit, R 2.15-2, I can no longer source
>> files for which the path contains the o-Umlaut (ö). As a historical
>> burden, my username is Grömping and contains that Umlaut (I wouldn't
>> have chosen it now, but decided to keep it for easy transfer). I have
>> had the difficulty that I couldn't use the R CMD tools on files with
>> Umlauts in the path (and solved it by having a dedicated directory
>> without umlaut in the path for that purpose), but so far (Windows XP and
>> earlier R versions) sourcing from within R worked fine.
>>
>> The issue is that the o-Umlaut (ö) is always replaced by a capital A
>> with tilde above together with a paragraph symbol (ö). Apparently, the
>> c3b6  used for the o-Umlaut in some encoding is translated back to the
>> ö. Can I somehow fix this? Or is it a bug?
>>
>
> I can't reproduce this.  The file d:/temp/Grömping/test.R contains
>
> cat("it worked!\n")
>
> I am on Windows 7 64bit, and I see the following in both 32 and 64 bit R:
>
> > source("d:/temp/Grömping/test.R")
> it worked!
>
> If I look at Encoding("d:/temp/Grömping/test.R") I get "latin1". If I run
>
> iconv("d:/temp/Grömping/test.R", "latin1", "UTF-8")
>
> I get a proper UTF-8 string, but if I run
>
> iconv("d:/temp/Grömping/test.R", "latin1", "utf8")
>
> I get a string with encoding marked as "unknown", and displayed as
>
> [1] "d:/temp/Grömping/test.R"
>
> So perhaps you or we have used an unofficial name of the UTF-8 
> encoding in some conversion.  We need to know exactly what you did to 
> pursue this.
>
> Duncan Murdoch
>
>
> Duncan Murdoch