excel_numeric_to_date converts incorrectly for dates between 1900-01-01 through 1900-02-28

[aetheogamout]

Because of the 1900-02-29 bug in how excel represents dates in windows, representation of dates from 1900-01-01 and 1900-02-28 effectively use a different origin than dates 1900-03-01 on.

The result of this is that dates produced by excel_numeric_to_date(1:59) are off by one day from the date as shown in excel:

in R:
excel_numeric_to_date(1)
[1] "1899-12-31"

in a modern version of excel convert the number 1 to a date:

=TEXT(1, "yyyy-mm-dd")
1900-01-01

[billdenney]

You're right. This was fixed in #422, and it will be included in the next janitor release. If you need to use it now, you can do remotes::install_github("sfirke/janitor") to get the current development version.

As an aside (see #423), excel date handling before 1900-03-01 is poor (searching the internet for "excel older dates" and similar will show many issues).

[sfirke]

Maybe a sign that it's time for a CRAN release 🤔

…

On Fri, Apr 15, 2022, 10:51 AM Bill Denney ***@***.***> wrote: Closed #478 <#478>. — Reply to this email directly, view it on GitHub <#478 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZYDECTRXNOLIK4N6L4FXDVFF7ADANCNFSM5TPRVXOQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[aetheogamout]

Thanks! and apologies - I misread what #422 is fixing

[billdenney]

@aetheogamout , No worries at all. Excel dates present all sorts of unique problems.

[billdenney]

@sfirke, if it's CRAN release time, I'm gonna make a couple more PRs. Hopefully, I can do those later today.