Skip to content

Errors and OS Differences with as.POSIXct and as.POSIXlt

4 messages · Bill Denney, Jeff Newmiller, Rui Barradas

Original
Bill Denney
# 
Hello,



Dates created with as.POSIXct differ between Windows/Mac and Linux.
Specifically this time that is during a gap when the hour does not exist
due to daylight savings time:



as.POSIXct("2018-03-11 02:09:36", tz="America/New_York")



Gives on Windows:

[1] "2018-03-11 EST"

Gives on Linux (Ubuntu 20.04):

[1] "2018-03-11 01:09:36 EST"



Since the time does not exist, and I think that NA should be returned.



Another issue and difference is that with as.POSIXlt on Linux, the invalid
time is presented:



as.POSIXlt("2018-03-11 02:09", tz="America/New_York")



Gives on Windows:

[1] "2018-03-11 EST"

Gives on Linux:

[1] "2018-03-11 02:09:00 EDT"



(Note that the time provided on Linux does not exist due to daylight
savings time.)



I think that for any invalid time, the result should be the same as an
invalid date:  NA is returned.



What is the intended, appropriate time, and what is the best way to fix
this?



Thanks,



Bill
Jeff Newmiller
# 
This is as described in the documentation, due to OS differences, e.g
 [1].

[1] https://stat.ethz.ch/R-manual/R-devel/library/base/html/strptime.html
On January 18, 2021 5:56:11 PM PST, Bill Denney <wdenney at humanpredictions.com> wrote:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Rui Barradas
    

    

      
      #
    

  


  

      
Hello,

R 4.0.3 on Ubuntu 20.04, sessionInfo() below.

A fix is to use as.POSIXct instead:

rui at rui:~$ Rscript --vanilla -e 'as.POSIXlt("2018-03-11 02:09", 
tz="America/New_York")'
#[1] "2018-03-11 02:09:00 EDT"
rui at rui:~$ Rscript --vanilla -e 'as.POSIXct("2018-03-11 02:09", 
tz="America/New_York")'
#[1] "2018-03-11 01:09:00 EST"


rui at rui:~$ Rscript --vanilla -e 'sessionInfo()'
R version 4.0.3 (2020-10-10)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 20.04.1 LTS

Matrix products: default
BLAS:   /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.9.0
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.9.0

locale:
  [1] LC_CTYPE=pt_PT.UTF-8       LC_NUMERIC=C
  [3] LC_TIME=pt_PT.UTF-8        LC_COLLATE=pt_PT.UTF-8
  [5] LC_MONETARY=pt_PT.UTF-8    LC_MESSAGES=pt_PT.UTF-8
  [7] LC_PAPER=pt_PT.UTF-8       LC_NAME=C
  [9] LC_ADDRESS=C               LC_TELEPHONE=C
[11] LC_MEASUREMENT=pt_PT.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

loaded via a namespace (and not attached):
[1] compiler_4.0.3


Hope this helps,

Rui Barradas
?s 01:56 de 19/01/21, Bill Denney escreveu:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Bill Denney
    

    

      
      #
    

  


  

      
There are many caveats about OS specificity on the strptime help page, but
most of them have to do with formatting and fewer with validation.  My
reading of the strptime page indicates that conversion with as.POSIXct()
will validate daylight savings times which the examples I gave indicate it
is not validating.

as.POSIXct on windows gives the expected value based on the help (the date
part is valid, so it exists, but the time part is invalid, so it does not
exist).  On Linux, it makes an inaccurate conversion (as shown in my
original message and in Rui's).

If the results from as.POSIXct do not guarantee valid time conversion, only
approximate conversion, then that should be made clearer in the help page.
If valid time conversion is intended as a guarantee from as.POSIXct and
as.POSIXlt, then this is a bug.