- From: Tab Atkins Jr. via GitHub <sysbot+gh@w3.org>
- Date: Thu, 13 Mar 2025 23:06:26 +0000
- To: public-css-archive@w3.org
Ah, my mistake, I didn't realize the text had behavior for premultiplying polar spaces. Cool. I assumed the manual handling of the bonus alpha was a required step, not just a convenience to avoid invoking the full interpolation machinery just to do a trivial alpha adjustment to the result. > [example with using tiny equal %] So we would have a discontinuity at 0%. Which may be the best we can do. Yeah, I think it's the best *generic* behavior to do - just have some reasonably neutral default value that gets triggered by this corner case. After all, you can't rely on limit arguments for this - if the initial %s are 20% and 80%, and they gradually scale down to 0%, they'll *also* arrive at 0% at the same time, but with a totally different interpolated hue. So any assumption you make about mixing weights will be wrong for some examples. "Might as well make them all equally wrong" isn't always the best thing to do, of course, but when talking about it generically, like with dynamic-range-limit-mix(), it's not clear there's any better answer. Just let the "default" value fill in any missing %, and if all the specified %s are 0, then the default value is what you get, regardless of any fancier mixing mechanics you'd get with small non-zero %s. I presume for dynamic-range-limit-mix() we'd say the default value is the initial value of `dynamic-range-limit`. (Note <https://github.com/w3c/csswg-drafts/issues/6245#issuecomment-2661545284> where we're defining the `*-mix()` family more generically, and explicitly talk about a "neutral" element taking up the leftover %. We didn't mention this corner case, but it follows naturally, I think.) -- GitHub Notification of comment by tabatkins Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/11678#issuecomment-2722879250 using your GitHub account -- Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config
Received on Thursday, 13 March 2025 23:06:27 UTC