[R-sig-finance] (i) bug report in listFinCenter and (ii) question

Dirk Eddelbuettel edd at debian.org
Wed Oct 5 03:01:13 CEST 2005


On 4 October 2005 at 14:46, Parlamis Franklin wrote:
|      Hello all.
|      I am using the Mac OS X version of R (Aqua GUI 1.12, R Framework  
| 2.1.1).
|     (i) The function 'listFinCenter' in the fCalendar package does  
| not work.  Having checked the code of that function it appears the  
| problem is in the passing of the '*' regular expression pattern to  
| the R grep function.  R complains that this is not a valid regular  
| expression.  I am not sure whether this bug is the result of more  
| liberal regular expression matching on the Windows platform (on which  
| fCalendar has been exclusively tested) than on UNIX, or whether it  
| results from the fact that R, as of version 2.1.0, has switched its  
| standard for regular expression matching to glibc 2.3.3 from grep 2.4.2.
|      In any event, I have found that replacing the pattern '*' with  
| the pattern '.*' will do the trick (although I must admit the entire  
| grep call seems a bit superfluous to me--perhaps the function was  
| originally intended to do more than it currently does).

I think Diethelm listens in here, so he may pick this up. If all else fails
you can always send bug reports to the package author/maintainer :)

|     (ii) As we all know by now, fCalendar needs the TZ environment  
| variable to be set to 'GMT'.  However, as installed on my Mac, a  
| Sys.getenv() call returned no environment variable named TZ.  My  
| solution was to use Sys.putenv() in my .Rprofile file to initialize  
| and set the TZ variable to 'GMT'.  This stopped the program from  
| complaining, but I am a bit concerned that the solution will not be a  
| complete one.  My system clock is still set to my local time zone and  
| daylight savings rules, so that other programs in other environments  
| on my system know the time to be different than R.  Does anyone  
| foresee a problem with this workaround?  Is there a better way?

I think that is a fair solution as .Rprofile should only affect R, and
processes spawned by R, if any.  So it should _not_ spill over to other

We have the same issue on Debian where timezone behaviour is controlled by
the file /etc/timezone, and TZ is unset by default. I also wish the code in
fCalendar (and similarly in its and other packages using timezone info) was
smarter amd set a default as fallback.

Best regards, Dirk

Statistics: The (futile) attempt to offer certainty about uncertainty.
         -- Roger Koenker, 'Dictionary of Received Ideas of Statistics'

More information about the R-sig-finance mailing list