Thank you for all the feedback. I feel the safest plan is to ship the entire module at once. It also saves some work to implement two preferences to exclude the shape-outside: <image> value which we don't render in the first stage.
I'm implementing "shape-outside: <image>", and will do "shape-margin" after that. My gut feeling is that the entire module can be completed before Firefox 60, which is a cycle late than the two-stages plan. Ting-Yu On Thu, Nov 30, 2017 at 7:37 AM, L. David Baron <dba...@dbaron.org> wrote: > On Thursday 2017-11-30 08:21 +1100, Xidorn Quan wrote: > > On Thu, Nov 30, 2017, at 07:16 AM, Mats Palmgren wrote: > > > I think supporting only a subset of the valid values for a property > > > is problematic. > > > > > > CSS conformance rules says: > > > "[...] the user agent must parse the value according to the property > > > definition. This means that the user agent must accept all valid values > > > and must ignore declarations with invalid values." > > > https://www.w3.org/TR/CSS21/conform.html > > > > It simply means we cannot state we are conformant to that spec in stage > > 1. That's not something seriously problematic. > > > > > Shipping support for a subset of the properties in a spec might make > > > sense > > > in some cases, but since you say: > > > > > > > The downside of the two-stage roll-out could raise web-compat issues > > > > because Blink and Webkit already support the entire module. > > > > > > it's not probably not a good idea in this case, for web-compat reasons. > > > > I'm not too concerned about this, actually. It really depends on whether > > web-compat issues from lacking of <image> support is worse than not > > supporting the property at all. > > > > I would argue that it isn't the case. shape-outside is mostly a cosmetic > > feature, and lack of support would unlikely cause anything more than > > some undesired appearance. In that case, authors probably don't even > > bother to use a feature detection at all. shape-outside itself has a > > popularity of 0.49% on Chrome Platform Status, which seems to be > > reasonably high, and I would be surprised if majority of that is using > > <image> rather than the simpler shape functions. > > > > Because of that, I don't really think web-compat would be a big issue > > for shipping this in two stages. > > I agree with Xidorn's conclusions. I'm skeptical of one of the > arguments, though, since I think "mostly a cosmetic feature" can be > said about many things that cause webpages to be unusable when they > don't work correctly. > > I think if the remainder of the feature can be done soon then it may > make sense to wait to ship the whole thing together, but if there's > something that makes the image part hard or unlikely to happen soon, > then it's reasonable to ship the part we have done, as long as the > part that we parse matches the part that we support. > > -David > > -- > π L. David Baron http://dbaron.org/ π > π’ Mozilla https://www.mozilla.org/ π > Before I built a wall I'd ask to know > What I was walling in or walling out, > And to whom I was like to give offense. > - Robert Frost, Mending Wall (1914) > > _______________________________________________ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform > > _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
