[R] No "number of days" format for 'difftime'?

[Martin Maechler]

>>>>> Loris Bennett <loris.bennett at fu-berlin.de>
>>>>>     on Mon, 8 Aug 2016 14:12:47 +0200 writes:

    > Loris Bennett <loris.bennett at fu-berlin.de> writes:
    >> Thanks for the link, John.  However, there is a hyphen missing.  It
    >> should be:
    >> 
    >> http://www.dummies.com/how-to/content/how-to-read-errors-and-warnings-in-r.html
    >> 
    >> Appropriately, with the correct URL we find the too often forgotten
    >> pearl of wisdom:
    >> 
    >> "Chances are, you just typed something wrong there."
    >> 
    >> I think I need this on my coffee cup.
    >> 
    >> Cheers,
    >> 
    >> Loris

    > Continuing the topic for my future self and others equally poorly versed
    > in The Art and Dark Science of Interpreting R Error Messages, if I have
    > the following in the file "my_data"

    > 1094165      2016-07-24T09:40:02 13-23:03:28          1  COMPLETED 
    > 1112076      2016-08-01T14:45:49 6-13:26:15          1  COMPLETED 

    > and do

    >> d <- read.table("my_data")
    >> colnames(d) <- c("jobid","start","elapsed","alloccpus","state")
    >> df <- transform(d,start = as.POSIXct(start,format="%Y-%m-%dT%H:%M:%S"),elapsed = as.difftime(elapsed,format="%d-%H:%M:%S"))

    > I get the following:

    > Error in as.difftime(elapsed, format = "%d-%H:%M:%S") : 
    > 'tim' is not character or numeric

Well, let me argue that you should have found this to be a *helpful*
error message. You are no complete beginner anymore, right,
so

1) the error is in your use of  as.difftime().

2) ?as.difftime  or  str(difftime)
   both clearly indicate that  'tim' is the first argument of as.difftime,

and I really do wonder why you continued with the infamous
"trial-and-error programming technique" instead of reading or at
least quickly browsing the relevant reference, i.e., help page

Martin


    > Remembering that if something is not what you think it is, it is
    > probably a factor, I find that

    >> d <- read.table(data_file,stringsAsFactors=FALSE)

    > causes the error to go away.

    > So what's all this 'tim' business?  A quick squint at the source code of
    > datetime.R reveals the following line:

    > if (!is.numeric(tim)) stop("'tim' is not character or numeric")

    > So the error message tells me something about the arbitrary name of the
    > variable 'tim', which could also have been 'tom', 'dick', or 'harriet',
    > but nothing about the value.

    > So follow Rolf Turner's fortune(350) advice, then look at the source
    > code, and then realise that you weren't being totally dopey in not
    > understanding the error message.


    > Loris

    > -- 
    > Dr. Loris Bennett (Mr.)
    > ZEDAT, Freie Universität Berlin         Email loris.bennett at fu-berlin.de