Fix `value-no-vendor-prefix` false negatives about stylelint HOT 14 CLOSED

please create another issue

Done.

Show a warning if…

I will only do it if the value starts with a known prefix.

Update stylelint-config-standard to rewrite the string values to regex ones for value-no-vendor-prefix
After the processes completes, we can fix false negatives reported in this issue. Any thoughts?

stylelint-config-standard will have to be updated twice: once for string to regex and once for the "/^-apple-system$/" addition. I would prefer to wait for both issues to be fixed on stylelint/stylelint to move on with the PR.

from stylelint.

Agree. We'd like to release stylelint and stylelint-config-standard only once.

from stylelint.

@Mouvedia Thanks for the report. Supporting the -apple- prefix sounds good to me. 👍🏼

And, adding regex support to the ignoreValues secondary option of value-no-vendor-prefix seems no problem. 👍🏼

Just to be sure, I have a few questions:

• Adding -apple- to reference/prefixes.mjs will affect selector-no-vendor-prefix, but are there any -apple- selectors? And, if there are (e.g. :-apple-foo-bar), will this addition fix also selector-no-vendor-prefix false negatives?

  • stylelint/lib/rules/selector-no-vendor-prefix/index.mjs

    Line 2 in 9959110

    import hasPrefix from '../../utils/hasPrefix.mjs';

• You suggest adding also -konq- and -khtml-, but what is rationale to add them, too?

from stylelint.

…but are there any -apple- selectors?

If you are talking about properties there are 3 AFAIK:

• -apple-dashboard-region
• -apple-line-clamp
• -apple-text-size-adjust

…will this addition fix also selector-no-vendor-prefix false negatives?

It will but it's not problematic. The problem is that this fix requires to update the ignoreValues option of value-no-vendor-prefix. It's similar to #7232 in the sense that a fix must be accompanied by its cure or it would be a downgrade.
And depending on our policies, updating ['string'] might require to wait for the next major.

There are 2 alternatives:

1. technically we could have 2 export for each rule (only change selector-no-vendor-prefix before v17)
2. add ["/regex/", /regex/] and wait for v17 to update ['string'] (make it exact match)

I vote for 2 because you can do exact match using regexes too. If so a new issue will have to be created akin to #7542.

tl;dr: the fix for selector-no-vendor-prefix is uncomplicated, for value-no-vendor-prefix it requires to update stylelint-config-standard and its secondary option.

…but what is rationale to add them, too?

Adding all the prefixes that ever existed makes sense.
At least that's what I would expect from *-no-vendor-prefix rules.

from stylelint.

Adding all the prefixes that ever existed makes sense.
At least that's what I would expect from *-no-vendor-prefix rules.

Okay. For -konq- and -khtml-, are there any resources? If yes, it seems good to leave a comment with links to such resources in reference/prefixes.mjs.

EDIT: Because many people may not know the -apple-, -konq-, and -khtml- prefixes.

from stylelint.

And depending on our policies, updating ['string'] might require to wait for the next major.

Just to confirm, does this update mean the following in my previous comment (#7624 (comment))? 👇🏼

adding regex support to the ignoreValues secondary option of value-no-vendor-prefix

from stylelint.

…are there any resources?

Yes, Iv linked to it in #7016 (review)

…it seems good to leave a comment with links to such resources in reference/prefixes.mjs.

👍

And depending on our policies, updating ['string'] might require to wait for the next major.

Just to confirm, does this update mean the following in my previous comment (#7624 (comment))? 👇🏼

adding regex support to the ignoreValues secondary option of value-no-vendor-prefix

If I did the PR, I would add support for ["/regex/", /regex/] without updating ["string"].
i.e. "a new issue will have to be created akin to #7542" to cover the breaking change

from stylelint.

Sorry, I can't still understand what you mean... It seems that it's not breaking to change ignoreValues: ["string"] to ignoreValues: ["/regex/", /regex/, "string"] 🤔

from stylelint.

It depends on how you proceed :)

Currently ["string"] picks up prefixes by default.
e.g. ['foo'] will match -webkit-foo
That's undesirable and unexpected and ["/regex/", /regex/] won't work that way.
e.g. stylelint-config-standard currently is ignoring more than it should
see stylelint/stylelint-config-standard#261 (comment)

So until the next major there will be inconsistencies between ["/regex/", /regex/] and ["string"].
Is that clearer?

from stylelint.

Ah, I now understand the problem here. 😅

The current ignoreValues option of value-no-vendor-prefix works on unprefixed values, but we'd like to support regex for a whole value (e.g. /^-apple-system/).

Indeed, this may a breaking change. So, cannot we add the regex support, keeping compatibility? 🤔

E.g.

 if (optionsMatches(secondaryOptions, 'ignoreValues', vendor.unprefixed(value))) { 
 	return; 
 }

+if (optionsMatches(secondaryOptions, 'ignoreValues', value)) { 
+	return; 
+}

from stylelint.

You can probably parse ignoreValues values.
Anyhow that's up to the implementor to decide.

from stylelint.

You can probably parse ignoreValues values.

Hum, I'd like to avoid parsing the option values since it seems too complicated... 🤔

from stylelint.

@ybiquitous Should I create another issue for ["/regex/", /regex/]?
If not is the secondary option update ready to be implemented?

from stylelint.

@Mouvedia Yes, please create another issue. For the secondary option, I guess:

• Add regex support. Specified regex should match a whole value rather than an unprefixed value. E.g. ignoreValues: [/^-apple-/] should match -apple-system or -apple-foo etc.
• Show a warning if a string is included in ignoreValues. E.g. String values in "ignoredValues" for "value-no-vendor-prefix" will match a whole declaration value rather than a unprefixed value in the next major version. If you want to keep the current behavior, please consider using a regex
• Update stylelint-config-standard to rewrite the string values to regex ones for value-no-vendor-prefix

After the processes completes, we can fix false negatives reported in this issue. Any thoughts?

from stylelint.