Closed brownag closed 1 year ago
Just got our two-weeks notice from CRAN that this needs to be fixed.
Dang. I don't have time to figure this out. Any suggestions other than removing the test?
This has now been fixed: tested locally on R 4.2.1 and R-devel, and run through winbuilder to confirm it is without error.
As I expected it was an issue with the specific timezone that these checks were being run. There are new tests that check the related problems locally... but this is nearly 600 timezones--so now after having verified they work on CRAN/R-devel I skip them. There are several where the time difference calculation will be plus or minus 1 hour depending on the start/end points... so that also needed to be accommodated in the conversion of difftime to integer.
I made a couple other changes as well:
1) better support for alternate WY end
date
2) changed default time zone to "UTC".
I think standardizing on UTC for these (and related) functions will make it easier to do localization if/when, for example, we add timezone metadata to the Henry DB or make use of it for SCAN data integration.
Thanks, this is a big help. Switching to UTC for computation was a clever idea. This is a good reminder to get the 'tz' column added to Henry Mount DB.
waterDayYear()
failing again (https://github.com/ncss-tech/soilDB/pull/223, https://github.com/ncss-tech/soilDB/pull/237) on CRAN for several R-devel architectures.This may not be specifically related to R-devel changes, as our regular tests pass on GHA, but rather likely related to the locale the tests are being run in combined with recent R-devel changes to date/time related functions
R-devel related changes: