The validation to number step is wrong

[cyfung1031]

@var range      param-1                 "Param 1"                               [12, 0, 36, 4, 'px']

Why I cannot set it to zero? When it become zero, it is just 12px
[updated: 2023.03.10 - I can no longer reproduce this issue; the default value issue can be closed.]

@var range      param-2                 "Param 2"                               [15, 3, 39, 4, 'px']

Why this is not allowed?

[eight04]

@tophf

Currently, we always assume the number is a multiple of step, which doesn't match browser's input[number]:

Only values which are equal to the basis for stepping (min if specified, value otherwise, ...) are valid
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/input/number

usercss-meta/lib/parse.js

Line 339 in b5d6e70

.some(n => n != null && !isMultipleOf(n, result.step))

Which should be changed to:

      .some(n => n != null && !canStepTo(basis, n, result.step))

However, this will be a breaking change, that value won't be a multiple of step anymore. WDYT?

[tophf]

The article says

When the data entered by the user doesn't adhere to the stepping configuration, the user agent may round to the nearest valid value, preferring numbers in the positive direction when there are two equally close options

so I think we should just forcefully snap min, max, default to the nearest step-multiple and emit a warning.

[eight04]

When the data entered by the user doesn't adhere to the stepping configuration, the user agent may round to the nearest valid value, preferring numbers in the positive direction when there are two equally close options

so I think we should just forcefully snap min, max, default to the nearest step-multiple and emit a warning.

I think it refers to user-input value, especially on a range input. The value will stick to nearest snap point automatically.

The issue in the OP is that, the following input is valid:

<input type=number value=1 min=1 max=5 step=2>

But the following code is not valid in usercss-meta:

@var number foo "Foo" [1, 1, 5, 2]

[tophf]

So the correct check should be value - min is a multiple of step, and if our check was wrong it doesn't matter that fixing it will be a breaking change. I expect most styles won't be affected because they were already using min/max that are a multiple of step.