Closed weierophinney closed 9 years ago
Following the reasoning of to restore the NonEmpty
behavior the correct thing could be inject NonEmpty validator instead setErrorMessage
Following the reasoning of to restore the NonEmpty behavior the correct thing could be inject NonEmpty validator instead setErrorMessage
I consider this a stop-gap, as we'll be removing it (as we've deprecated this use case with earlier patches). As such, we're just trying to mimic previous behavior at this point.
This part is not deprecated (still smelling of new) so this could be reused in future developments.
The API suggested should be reused on v3.
I found this way does not allow customization through input specifications.
Per #28, since 2.4, we've not been returning error messages in the expected format when a value is required but no value (or an empty value) was provided for the input.
This patch builds on #60, and update the test expectations to those reported in #28.