There is something wrong, and a regression has been caught by those
tests:
If an invalid date is passed from the add item form, the app now
crashes.
Before:
* if the date was completely invalid, the field was blanked
silently
* DateTime::Format::MySQL was used to convert dates, and it's not
strict at all. For instance, what happened in the selenium tests for
dateaccessionned: %Y-%m-%d was prefilled by the framework plugin, then
the biblionumber was added, we ended with something like (eg for today)
2020-03-234242 (with biblionumber=4242). DateTime::Format::MySQL
converts that to 2020-03-23
We must deal with invalid dates, but I do not think it is good to add it
back to Koha::Item->store, we will prefer to raise the error to the end
user, saying that something went wrong (and more specifically the
dates).
The (ugly) trick was in C4::Items::_mod_item_dates
Signed-off-by: Martin Renvoize <martin.renvoize@ptfs-europe.com>
Jonathan Druart
757ca57bfd
