Understand, but in certain type of application they would be very useful. For example importing data from console or in complex models.
At the moment one correct way i found is to create model methods named for example “checkWarningA” , “checkWarningB” ecc…
And in every point of the controllers that i run the model validation , i must remember to call every “checkWarning” function and get the warning message from the model through constants.
But in this way i can not use the standard core validators .
If we standardize the warnings , must be useful and simple to add a property to the base validator that would make it “soft”, adding a warning instead of an error.
In this way we could simply manage the warnings using all the logic already present in the model, like the rules() method and the scenarios.
I don’t think that the warnings would increment the “work” of the model. Because if you do not use them it will not change anything in the model validation process.
If with “work” did you mean to add new methods to the model , we can reuse the error functions with a specific parameter, but personally i prefer the first way .
I would be happy to hear other opinions or ideas