As Dirk mentioned earlier, we had quite a few people from the Web Platform team at Adobe attend TPAC this year. Here’s my summary of what happened for the Regions and Exclusions specifications at the CSS Working Group meeting.
Rossen Atanassov from Microsoft led this discussion. The working group resolved to close several old issues. The first was whether the wrap-through property is necessary, and we demonstrated the use cases where it is required. The second was about how floats and exclusions interact. Since the issue was raised, we’ve added a section to the specification describing how they work together. And the third issue was closed because we’ve added the Exclusions Processing Model section that describes how wrapping contexts work.
We also discussed the shrink-to-fit issue, which is how shrink-to-fit should be calculated when laying out content inside (for example) a circle. The intent is to allow the shape to be just large enough to fit its content without causing overflow. Rossen has proposed a two-pass algorithm that will be added to the specification soon, and I expect we’ll have some spirited discussion on the www-style mailing list once that happens.
In the Regions discussion, we closed two old issues. The first was about auto-generating regions boxes in order to accommodate variably-sized content. Instead of box generation, the spec now has auto-height regions and a processing model that determines how the boxes respond to more or less content. The second was about Regions creating stacking contexts. The working group decided that they will create stacking contexts, as that simplifies the implementation.
We also argued over the sticking point of region boxes being defined by HTML elements. The group has many opinions (it’s good, it’s theoretically bad but practically useful, it’s completely unacceptable) and has not come to consensus on the issue. One interesting point came up in discussing the Regions implementation in IE10. They use HTML elements to define regions, but have them in a separate file from the content. This is comparable to one way that <template> elements can be defined in Web Components. Two of the people that find Regions elements unacceptable said that separating layout from content in this way might be OK. So I have a new proposal to use Web Components ideas in the CSS Regions examples.
We made some good progress on both specs at this meeting, and in the upcoming months we will be addressing more issues both in the specifications and in our implementations. Hopefully we will get both specifications to zero issues soon, and we’re working on test suites to prove implementation progress.