R-devel changes from last week's dev day lead to datetime check bug (original) (raw)

Issue

R Contributors had a Dev Day last week in conjunction with LatinR. @hturner and I worked on an old datetime issue which we got resolved and merged to R-devel thanks to @mmaechler. Now, of course, no small good deed goes unpunished as the buglet just found here reveals that the fix in R-devel may have been suboptimal. I just sent a suggested fix to the Bugzilla bug tracker on that issue which is here.

Suggested Fix

We can work around this by supplying a slightly different fixed format string:

diff --git a/inst/tinytest/test_date.R b/inst/tinytest/test_date.R index 84891ae5..a24e2668 100644 --- a/inst/tinytest/test_date.R +++ b/inst/tinytest/test_date.R @@ -194,7 +194,7 @@ expect_equal(Datetime_format(d,"%Y-%m-%d %H:%M:%S"), format(d, "%Y-%m-%d %H:%M:%OS"), info="Datetime.formating.default") expect_equal(Datetime_format(d, "%Y/%m/%d %H:%M:%S"),

•         format(d, "%Y/%m/%d %H:%M:%OS"),

•         format(d, "%Y/%m/%d %H:%M:%OS6"),
          info="Datetime.formating.given.format")

expect_equal(Datetime_ostream(d), format(d, "%Y-%m-%d %H:%M:%OS"),

which avoids the isssue tickled and will do no harm whether my additional change gets accepted or not.