Custom validation error doesn't hide on good result

[shishmak]

Prerequisites

• I have searched the existing issues
• I understand that providing a SSCCE example is tremendously useful to the maintainers.
• I have read the documentation
• Ideally, I'm providing a sample JSFiddle, Codesandbox.io or preferably a shared playground link demonstrating the issue.

What theme are you using?

utils

Version

5.x

Current Behavior

Hello,
I need to have custom validation depense with 2 fields
Previosly I use rjsf 3.x and all works good. I upgrade project to 5.x and find bug (as I think).

I have a form with a dependency "Test" <= "Test2". I applied a custom validation where I specified this dependency.
const validate = (formData: any, errors: FormValidation) => { if (formData["Test"] > formData["Test2"]) errors["Test"]?.addError("Test should be LE than Test2"); return errors; };
And I noticed a problem:
Let's say initially I have Test = 10 and Test2 = 15

If I set Test2 = 5, then an error will appear near Test. This is the correct behavior

If I return Test2 = 15, then the error will not disappear. And this looks like a bug, because the values ​​​​Test (10) and Test2 (15) satisfy the condition "Test" <= "Test2"

example of my code:
https://codesandbox.io/p/sandbox/yyrxsp

Expected Behavior

If no validation errors are found, then no errors should be shown.

This error should not be there,

because according to the code the condition = false
const validate = (formData: any, errors: FormValidation) => { if (formData["Test"] > formData["Test2"]) errors["Test"]?.addError("Test should be LE than Test2"); return errors; };
Test < Test2 (10 < 15)

Steps To Reproduce

https://codesandbox.io/p/sandbox/yyrxsp

1. set Test 2 = 5 (an error will appear near "Test 1")
2. set Test 2 = 15 (the error near "Test 1" will not disappear)

Environment

- OS: windows 10
- Node: v20.11.0
- npm: 10.3.0

Anything else?

No response

[heath-freenome]

@shishmak First of all, you are using MobX incorrectly. In your sandbox I am seeing the following warnings:

[MobX] Since strict-mode is enabled, changing (observed) observable values without using an action is not allowed. Tried to modify: ObservableObject@2.jsonScheme 
    at eval (https://s9pdhk.csb.app/node_modules/mobx-react-lite/es/observer.js:43:47)
[MobX] Since strict-mode is enabled, changing (observed) observable values without using an action is not allowed. Tried to modify: ObservableObject@2.uiScheme 
    at eval (https://s9pdhk.csb.app/node_modules/mobx-react-lite/es/observer.js:43:47)
[MobX] Since strict-mode is enabled, changing (observed) observable values without using an action is not allowed. Tried to modify: ObservableObject@2.formData 
    at eval (https://s9pdhk.csb.app/node_modules/mobx-react-lite/es/observer.js:43:47)

When doing a console.log of your formData as I type, I don't see it updating so perhaps this is leaving the form in a confused state.

Perhaps once you fix things with MobX the issue may be fixed?

[shishmak]

@heath-freenome I fixed the mobX errors and added an example on React States. The error remained in both places

[donalddalton]

@shishmak @heath-freenome any update on this? I am running into the same issue after attempting to upgrade RJSF where errors inserted via a custom validator remain in errorSchema despite being cleared from the form errors state. I see this new behavior starting in version 5.20.0 so possibly related to #4188? Thanks!

[heath-freenome]

@donalddalton Can you provide an example in a code sandbox (or shared playground) that we can use to determine whether your issue is a regression for #4188 ?

[donalddalton]

@heath-freenome sure thing! Here is a crude example recapitulating the scenario @shishmak described above https://codesandbox.io/p/sandbox/rjsf-example-4m3stz. This example uses version 5.20.0/MUI/ajv6. You can reproduce by:

• start by setting Test=1, Test2=2

• changing Test -> 10 shows the expected custom validation error in the error list and inline under the Test field

• changing Test2 -> 20 removes the error from the error list but it remains inline under the Test field

[heath-freenome]

@abdalla-rko Hey Appie, let's chat in Discord about this next week. You introduced a regression with your #4188 changes