fastest-validator
fastest-validator copied to clipboard
icebob
Custom validator regression with multiple rules
#272 seems to have introduced a regression when using custom validations and multiple rules between 1.11.1 and 1.12.0.
If there are multiple rules, and an earlier rules fails while a later rule passes, the validation will still fail.
In this example codesandbox, you'll see a custom rule for a simple "even number" example. When using 1.12.0, the "Array passing" rule is failing, even though it should pass. If you change the version of fastest-validator to 1.11.1 then this same validation will be successful.
I debugged through a bit and I believe the problem is in makeCustomValidator method. Errors that come back from custom validators are pushed on to the main errors array. This wasn't a problem in 1.11.1 because the multi validator code used to confirm if the number of errors had changed between each validation, but in 1.12.0 it is instead using different variables for the results of each validation and conditionally pushing those results on to the errors array. However, because the custom validator is itself pushing on to the errors array, the multi validator's behavior is not occurring as expected.
@shawnmcknight could you check that 290 solves your issue as well? It's not released you, you should install the master branch
@0x0a0d @icebob Unfortunately, no I think its still a problem. I forked the codesandbox and installed from GitHub instead of npm and the "Array passing" test example using an array is still failing.
I think it's codesandbox problem, I tried with master branch and the test worked as expected
@0x0a0d @icebob Hmm, yeah, not sure why the codesandbox is still failing. Maybe codesandbox doesn't properly allow installing from GitHub? Anyway, I had a failing unit test on my local repo with 1.12.0 and just installed the master branch version locally. The unit test is no longer failing, so I do think this resolves the issue. 👍