CSS WG Blog front page
Resolutions 2007-11-05 2007-11-06 Boston Part III
Daniel Glazman brought up the 1998
List of suggested extensions to CSS
and how we haven’t been aligning our work with designer needs, even with the
feedback that we have already. This sparked a long conversation, which I don’t
think I can summarize, but I’ll try to pull out some of the points made here.
Note that these points aren’t representing WG consensus, just some ideas that
came up in the conversation: partly quotes, partly paraphrasing, partly
summarizing.
- We have not done a good job of considering user feedback.
- Some of the requests in the feedback were not pursuable for various reasons
back when the feedback was given; others were simply not acted upon.
- We have some feedback in the “List of suggested extensions to CSS”.
We should examine this list and see a) if the issues are still relevant and b) if we are ready to solve them
- We’ve been dominated by browser makers and need more user input, but
we are not set up in a way that makes it easy to collect this information.
- CSS is driven forward both by pull from features specified by the CSSWG
and by push features developed by implementors.
- Some ideas fly; others do not; we need to piece together solutions from the best parts.
- Where there is no champion for a spec, progress on that spec ceases.
- Interoperability is important to web designers.
- Good test suites drive interoperability. Wrong test suites drive
interoperability on wrong behavior.
- We should solicit and respond to user feedback.
- We should use the CSS blog to collect feedback — but we don’t have that
capability now — [various technical suggestions on what to do]
- We are happy for the CSS Eleven or any other group to gather feedback
and process/organize it into something we can act on.
- We should document personas, use cases for what we’re trying to do.
Someone diagrams our efforts into the following sections:
- Readability (typography, fine layout)
- Printing
- UI / coarse layout
- Internationalization
- Foundation (e.g. box model)
- Programmability (making desired results easy to declare)
See also Kevin Lawver’s Web Standards’ Three Buckets of Pain.
Resolutions 2007-11-05 2007-11-06 Boston Part II
Marquee
(See overflow section in CSS3 Box Module draft.)
-
Resolved: Rename marquee-loop to marquee-play-count.
Rationale: Avoids confusion over whether marquee-loop: 1 means play once or play twice. This was a major point of confusion for the HTML5 loop attribute, which was renamed playcount for this reason.
-
Noted: Need clarifications for behavior of marquee-style values.
-
Noted: Computed value section of marquee-direction doesn’t match values.
-
Noted: Note about alternate under marquee-direction definition could be interpreted incorrectly and should be fixed by saying “opposite of the direction given by marquee-direction” or somesuch instead of “opposite direction”.
-
Resolved: Remove overflow-style-x and overflow-style-y in favor of overflow-style: auto | marquee-line | marquee-block where marquee-line scrolls parallel to the line boxes and marquee-block scrolls perpendicular to them.
Rationale: Combining this into one property avoids senseless combinations like marquee in both directions, and also works the same way for both vertical and horizontal text.
Box Module and Width Keywords
-
Resolved: Keep display: compact in CSS Box Module Level 3.
Noted: David Baron sent many comments on display: compact; they should be dug up and addressed.
-
Resolved: Add to width, min-width, max-width, height, min-height, and max-height keywords from Section 3 of David Baron’s intrinsic sizing document, but rename fill to available, min-intrinsic to min-content, intrinsic to max-content, and shrink-wrap to fit-content. Define that all 4 values should act as the property’s initial value when they are specified on the inline progression dimension. (Add also examples to demonstrate their use.)
CSS2.1 Issues
Media Queries
-
Resolved: Define in CSS3 Paged Media that @page rules are allowed inside @media blocks.
-
Resolved: all and can be dropped from a media query, e.g. @media (color) {} is valid.
-
Resolved: Unknown/unsupported media queries are ignored, but the other comma-separated parts can still be used.
-
Resolved:Dropping unknown parts of a media query such that all parts are dropped means you end up with none, not all.
Media Queries will go back to Last Call for these and other changes.
Rotation and Transformation
Discussed use cases, SVG-like transform syntax, and centers of rotation.
-
Consensus: We need keywords for pegging the center of rotation at, e.g. the baseline of an inline element, or at the center of any other element.
-
Noted: For the above reason and for usability, we need a way to set the center of rotation explicitly.
-
Noted: A diagonal keyword for the angle of rotation could introduce circularity if it affects layout.
-
Noted: Separate properties to trigger transform behaviors could introduce cascade problems.
-
Resolved: transform property should trigger a new block formatting context. A none value is needed as the default.
-
Noted: Importing SVG syntax directly doesn’t work because it’s incompatible with CSS tokenization and other syntactic conventions. Specific problems include: CSS can’t accept spaces before the opening ( of functional notation, functions need to take lengths not numbers, and functional notation should require commas.
Discuss on www-style
Resolutions 2007-11-05 2007-11-06 Boston Part I
The W3C has an annual Technical Plenary, during which all its working groups have face-to-face meetings in the same place. This gives groups a chance to have joint meetings to work out cross-working-group issues. Due to the way scheduling was handled this year, we only got two days of meetings whereas we usually have three.
CSS and MathML Joint Meeting
The MathML Working Group is working on ways to display MathML with CSS. Our discussions resulted in the MathML WG posting a list of Prioritized Requests for CSS3.
CSS Validator
The CSS Validator team requested a better definition of what a valid CSS document is in Level 3 and clearer indications that REC-CSS1 and REC-CSS2 are obsoleted.
Tables
We reviewed a new version of the table width calculation algorithm document. This text will be incorporated into a new CSS3 Tables module soon.
CSS3 Multi-Col
Agreed to publish an LC draft (after vertical text issues are addressed) with an extended public comment period to invite more feedback.
Celebrating 10 Years with Style
In December 2006, W3C celebrated that the CSS1 Recommendation was 10
years old. Part of the festivities was to launch a site to record some
of the history, seminal articles, a hall of fame, and a gallery of
CSS-based designs.
We invited people to submit their favorite CSS pages and designs for
consideration. The idea was that Bert Bos and myself would admire the
fine works of arts and add our favorites to the gallery; perhaps some
would even go into the hall of fame.
Due to, ahem, some technical problems with, ahem, the email system of
a consortium near and dear to us, the messages sent for consideration
bounced during most of the year. We apologize for this.
We therefore humbly ask people to resubmit their CSS pages and design
for consideration. The site will be frozen before the end of the year
and this is your chance to help define a CSS time capsule.
Please send your your contribution to: css10@w3.org
Resolutions 2007-10-16, 2007-10-23, 2007-10-30
Other notes:
- Most discussion was on font formats and IPR this month.
- Some brief discussions of other topics such as rotation and media queries.
The CSS Recommendation Track
In the 2007 CSS Snapshot I briefly explained the W3C Process, how it relates to the CSS Working Group’s experience, and how its stages do not correspond very well to stability levels in our work. In this post I want to outline the stages that I’ve observed in our work of defining CSS and point out some examples from our active repetoire. (As with the official W3C process stages, a spec can move backward in response to feedback as well as forward.)
- Exploring
- In this stage the spec is often incomplete, possibly changing greatly between drafts, and possibly including many features that will be dropped as the module matures. Examples of specs at this stage include:
- Rewriting
- Some modules enter this stage, where large parts of the spec are rewritten. Examples of specs at this stage include:
- Revising
- At this point the spec is mostly complete and the scope of its functionality is well-defined, but the spec still needs several cycles of publishing, review, and revision to uncover issues and resolve them. Examples of specs in this state include:
- Refining
- At this point the spec is almost stable enough for CR, but still needs some well-defined changes from e.g. last-call comments, or general minor polishing. Specs in this state include:
- Call for Implementations
- At this point the WG believes the specification to be complete and precise enough to be implemented, and by transitioning it into the CR status has issued a call for implementations and test cases. Specs in this state include:
- Stable
- Although the test suite and implementation reports may not be done yet and there may still be a few minor issues left, at this point the WG has enough implementation experience that it considers the spec ready for wide use. Specs in this state include:
- Completed
- At this point the test suite and implementation reports are all done, the specification is officially a W3C Recommendation, and no changes are expected. None of our specs have reached this state yet.
New draft of Behavioral Extensions to CSS
We published a new draft of Behavioral Extensions to CSS a few days ago. It changed significantly from the previous draft published in 1999. The new draft defines a binding CSS property and :bound-element pseudo-class. The other features are now covered by the XBL draft which is being developed by the W3C Web Application Formats Working Group.
As for the features in the draft, the binding property allows you to link to resources in the XBL format and the :bound-element pseudo-class matches elements that have a binding.
Comments are welcome on the www-style@w3.org mailing list!
CSS2.2 and the 2007 CSS Snapshot
There was a lot of discussion earlier this year around Andy Budd’s proposal for a CSS2.2. The basic premise was that the Web needs a halfway point between CSS2 and The Complete CSS3 that is taking forever, so that the key features web developers need now can happen sooner. The structure of CSS3 is actually set up so that this can happen, but the CSS Working Group has realized that this is far from obvious to anyone outside the working group. So we’ve decided to publish a CSS2.2.
Except we’re not going to call it CSS2.2, because to the Working Group CSS2.2 means something completely different. (It means CSS Level 2 Revision 2, which, really, isn’t something anybody needs right now.) We’re also not going publish just one “CSS2.2”, but many—hopefully one a year. The first one is the Cascading Style Sheets (CSS) Snapshot 2007. Our goal in publishing this document is to explain the state of CSS, demonstrate the modular nature of CSS3, and make many of the tacit recommendations of the CSS Working Group explicit. This snapshot defines, as Markus puts it, The Book of CSS: it’s the introduction and table of contents for everything that would be part of the 2007 Edition of Cascading Style Sheets: The Definitive Spec.
If you’re interested in the development of the CSS specs, read the Snapshot. It’s very short (4 pages printed), and mostly explanation rather than technical definition. There’s nothing new or terribly exciting in it (expect more in 2008), but I hope this document makes it easier for people to understand the structure and development of CSS.
Resolutions 2007-10-02
-
Resolved: Publish a new draft of the Grid module soon to get away from * syntax
-
Resolved: Start work on next Paged Media spec, since we want to finish stabilizing CSS Paged Media Level 3 but also have a lot of ideas that need hashing out. These include (but are not limited to):
- A
:blank page selector for blank pages generated by forced page breaks.
- An
allow value for page breaking, to override a parent’s avoid.
- Repeating captions across pages, including with different text (e.g. the caption on the first page says “Table 1”; the caption on the next page says “Table 1 (cont.)”).
- better control of the
pages counter (skip pages, reset, etc.)
- top-of-page margin collapsing control
- fallback sizes for the ‘size’ property (e.g.
size: letter, A4;)
- some of what’s currently in GCPM.
- other proposals
-
Resolved: Since we now have public workspace on dev.w3.org, we resolve that at least two implementors must be interested in implementing a module before we publish an official W3C Working Draft on the TR page.
Other topics discussed:
Resolutions 2007-09 Beijing Part V: CSS3 Tables, Box, Grid, and Snapshot
Tables and Intrinsic Sizing
Discussion on www-style
No resolutions from this session, which took up the whole morning. It was a working session where the WG tried to come to a common understanding of automatic table layout’s column width calculation in (almost) all its gory detail. You can see some of the gory details in this discussion on www-style and more extensively in David Baron’s document and Microsoft’s document.
Microsoft’s team and Mozilla’s David Baron have each put together partially-intersecting discussions of table layout, which we hope to put together into detailed spec of how automatic table layout should work for the CSS3 Tables module. The goals are to spec what’s currently interoperable, and where there isn’t interoperability, to spec what is partially interoperable and sensible.
CSS Grid Layout
General discussion and some resolutions on syntax issues:
- Resolved: Change terminology from “natural grid” to “implicit grid”
- Resolved: Drop default grid and look up the containing block chain for a grid
- Resolved: Use ‘fr’ instead of ‘*’ for flex units (see HTML tables for *).
- Resolved: Use ‘repeat(…)’ to enclose repetition patterns.
image-orientation Value Parsing
-
Resolved: image-orientation accepts all angles, rounds to nearest increment of 90°, and spec says authors SHOULD NOT specify angles other than increments of 90°.
Rationale: Since a number followed by alphabetic is parsed as a length, we can’t allow 90deg but not 90.0deg. If we allow radians, then what matches? 3.14rad and 3.14159rad are both approximations of 180°, but they are neither equal to 180° nor equal to each other. We could either allow only degree measurements (which would be odd because all other angle properties accept all angle units, just like all other length properties accept all length units) or accept all angle values and simply round to the nearest 0°/90°/180°/270° angle. The WG, to avoid changing parsing behavior and because it seemed more intuitive, opted for the latter.
CSS Box Module Level 3
CSS Snapshot 2007
-
Resolved: The CSSWG will publish a document that collects together all the CSS specs that we consider stable, a profile that declares the scope and state of CSS at this point in time. It will include only specs that we consider stable in their definitions (i.e. we don’t expect significant changes) and for which we have enough implementation experience that we are sure of that stability.
-
Resolved: The 2007 version of this document will include
- CSS2.1
- Selectors
- CSS3 Color
- CSS Namespaces
-
Noted: The 2008 version of this document is likely to contain also CSS3 Paged Media and Media Queries, possibly also CSS Ruby and CSS3 Backgrounds and Borders depending on how those drafts progress.
Archives
Browse by date:
Browse by category:
![[Photo: group photo of the CSS working group in San Francisco]](/Style/CSS/csswg-2016.jpg)
Contact: