[R] library(fCalendar) timeDate("12.03.2005", format="%d.%m.%Y")

[Martin Becker]

Martin Maechler wrote:
>>>>>> "OL" == Ola Lindqvist <ola.lindqvist at nr.no>
>>>>>>     on Tue, 21 Aug 2007 14:32:19 +0200 writes:
>>>>>>             
>
>     OL> Thanks!
>     OL> Seems to work fine now!
>
> Well, for your example.
>
> But sorry to say, the patch breaks other cases.
>
> I'm investigating further
> (and will hopefully contribute to a new CRAN release of fCalendar
>  once Diethelm Wuertz is back from wherever; I've already made
>  more changes)
>
>   

Good to see experts investigating :-)

Sorry, I should have mentioned, at least, that I am not an Rmetrics 
developer, that I only took a quick glance at the code (focused on the 
reported problem), and that the patched version is neither official nor 
tested. Now I see that I completely messed up the case for "Date" input 
and sometimes .midnightStandard() was not working correct any more.

One further issue, which was brought to my attention by another R-user 
off list: I think, that all occurences of

  if (sum(lt$sec+lt$min+lt$hour) == 0) isoFormat = "%Y-%m-%d"

in R/3A-TimeDateClass.R could safely be replaced by

  if (sum(lt$sec+lt$min+lt$hour,na.rm=TRUE) == 0) isoFormat = "%Y-%m-%d"

in order to make fCalendar more NA-friendly. Currently, at least for 
some cases, a single NA entry in the input vector suffices to let 
timeDate() break down. Again, I am not sure, if this change breaks some 
special cases, or if breaking down in case of NAs is intended.

Nevertheless, I have built a new, unofficial and untested patched 
(windows-)version of fCalendar (same weblink), which makes the "Date" 
input case and .midnightStandard() working again (I hope so, at least), 
and incorporates  the (improved?) NA-handling. But of course: only use 
this version with care (and only if the official version doesn't work 
for your particular problem). Hopefully, we see an official update on 
CRAN soon :-)

Kind regards,

  Martin Becker


> Martin Maechler,
> ETH Zurich  [but different department than D.Wuertz]
>
>
>     OL> Martin Becker wrote:
>     >> Dear Ola,
>     >> 
>     >> I think you spotted a small bug in *package* fCalendar.
>     >> Explicit specification should prevent "autodetection" of the date 
>     >> format, which is not the case for fCalendar v251.70, instead 
>     >> autodetection is done at least once (twice, if actually appropriate).
>     >> With the following patch, things should work ok:
>     >> 
>     >> diff --recursive fCalendar.orig/R/3A-TimeDateClass.R 
>     >> fCalendar/R/3A-TimeDateClass.R
>     >> 433c433
>     >> <     charvec = format(strptime(charvec, .whichFormat(charvec)), 
>     >> isoFormat)
>     >> ---
>     >> >     charvec = format(strptime(charvec, format), isoFormat)
>     >> 
>     >> You did not provide the output of sessionInfo() (which you are asked 
>     >> for in the posting guide). If you are using Windows and don't know how 
>     >> to apply the patch, you can download a patched binary version here:
>     >> http://www.saar-gate.net/download/fCalendar_251.70.zip
>     >> 
>     >> Regards,
>     >> 
>     >> Martin
>     >> 
>     >> PS: Maybe r-sig-finance is more appropriate for questions concerning 
>     >> Rmetrics.
>     >> 
>     >> 
>     >> Ola Lindqvist wrote:
>     >>> Dear R users,
>     >>> I have problem with the library fCalendar.
>     >>> 
>     >>> I am not using the US standard format notations. It seems like it is 
>     >>> not possible to have different format than the US standards.
>     >>> Anyone how knows a way to go around this problem?
>     >>> 
>     >>> Here is the code I enter:
>     >>> myDate = "12.03.2005"
>     >>> timeDate(myDate, format = "%d.%m.%Y")
>     >>> 
>     >>> And I get following error message:
>     >>> Error in if (sum(lt$sec + lt$min + lt$hour) == 0) isoFormat = 
>     >>> "%Y-%m-%d" :
>     >>> missing value where TRUE/FALSE needed
>     >>> 
>     >>> Thanks,
>     >>> Ola
>     >>> 
>     >>> ______________________________________________
>     >>> R-help at stat.math.ethz.ch mailing list
>     >>> https://stat.ethz.ch/mailman/listinfo/r-help
>     >>> PLEASE do read the posting guide 
>     >>> http://www.R-project.org/posting-guide.html
>     >>> and provide commented, minimal, self-contained, reproducible code.
>     >>> 
>     >> 
>
>     OL> ______________________________________________
>     OL> R-help at stat.math.ethz.ch mailing list
>     OL> https://stat.ethz.ch/mailman/listinfo/r-help
>     OL> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>     OL> and provide commented, minimal, self-contained, reproducible code.
>