search

postcss / autoprefixer

Parse CSS and add vendor prefixes to rules by Can I Use
https://twitter.com/autoprefixer
MIT License
21.57k stars 1.25k forks source link

Threshold for dropping "mixed support" warnings? #1513

Closed knulpi closed 3 months ago

knulpi commented 3 months ago

I was wondering what the threshold is for a value to not be considered as mixed support anymore. Taking as example the 6 year-old mixed-support-warning for using start/end instead of flex-start/flex-end. The support for start/end is marked as 94% across browsers based on caniuse.com.

I personally would consider this not being a case of "mixed support" anymore, but is there any rule for this?

ai commented 3 months ago

It was a bigger problem a few years ago. But still is good advice since 6% is still more 6% for free.

If we have no reasons to remove it, I prefer to keep it.

knulpi commented 3 months ago

But as the CSS Box Alignment Module Level 3 is to be the new baseline for these values, wdyt about making the warning a bit more verbose?

Something like "'end' value might not be supported in older browsers, consider using legacy 'flex-end' for broader support (see https://caniuse.com/mdn-css_properties_justify-content_flex_context_start_end for supported browsers).

ai commented 3 months ago

We can remove the warning if you will present me rational reasons why end could be better than flex-end.

knulpi commented 3 months ago

I got new input on this. While I thought that the flex- versions are equivalents, they aren't actually. The behaviour is different, based on the flex-direction property, as can be seen in this codepen.

So this is even a better reason to reword/remove the warning, as this is not only a topic of mixed support, but also of different/unintended behavior.

ai commented 3 months ago

Good reason. I will disable warning today.

ai commented 3 months ago

Done https://github.com/postcss/autoprefixer/commit/fe7bae45970d38eee86c4d26e5f19fa100df3039

Released in 10.4.19.

Thanks for good arguments.