Should EPUB3 introduce prefixes to properties in CSS WD? (Part 1)
MURATA Makoto
available at:
http://code.google.com/p/epub-revision/wiki/Telcon20110413_Minutes
I volunteered to write a summary of this issue and list possible options.
1. Background
EPUB3 relies on many moving targets, which are listed at:
http://code.google.com/p/epub-revision/wiki/MovingTargetRiskManagement
In particular, EPUB3 uses some CSS working drafts. They are CSS Fonts
Module Level 3, CSS3 module: Lists, CSS3 ruby module, CSS3 Speech,
CSS3 Text, and CSS Writing Modes. Some (but not many) of the
properties in these specs are likely to be changed before the specs
become recommendations.
EPUB3 also uses CSS 2.1 (Proposed Rec), CSSMultiCol (Candidate Rec),
and CSS Namespaces (Candidate Rec). The properties in these specifications
are assumed to be stable, although this has not been always the case in the
past.
2. Cupertino Writing Party
This issue was discussed and Fantasai was requested to report back to
the CSS WG. See the beginning of "Summary of Outcomes"
(http://code.google.com/p/epub-revision/wiki/WritingParty201101Cupertino)
Review of this profile by community and general and CSS WG in
particular will happen during spring; during this phase
changes (-idpf prefixing in particular) may occur
3. CSS WG
This issue was extensively discussed in a teleconference on 4/14. The
minutes are available at:
http://lists.w3.org/Archives/Public/www-style/2011Apr/0429.html
Please read this minutes carefully. The biggest concern of the CSS WG
appears to be lock-in to the present working drafts. The CSS WG might
want to change the properties in the working drafts for good reasons
in the future and do not want to be blocked by the upcoming corpus of
EPUB publications.
Their conclusion is as follows:
> RESOLVED: Epub should use prefixed versions of properties that aren't
> yet in CR. No recommendations on what the prefixed property
> means.
But the minutes also show interesting observations:
> plinss: What I'm hearing dissent on is what the prefixed version means.
> plinss: They're their own working group. We can make recommendations.
> If they decide not to prefix, they don't prefix and the only
> thing we can do is get mad about it.
4. Maturity of CSS WDs
4.1 CSS Text
EPUB3 uses several (not all) properties from CSS3 Text. The syntax of
'text-trim' is very likely to be changed. The syntax of other
properties referenced by EPUB3 are much less likely to
change. Semantics of many properties will change for edge cases. We
might want to drop the text-trim property from EPUB CSS Profile.
4.2 CSS Writing Modes
EPUB3 uses several (not all) properties from CSS Writing Modes.
Syntactical changes are not expected, but semantics of many properties
will change for edge cases and one common one of auto-sizing when
switching writing modes.
4.3 CSS Fonts Module Level 3
EPUB3 uses several (not all) properties from CSS Fonts Module Level 3.
The bits of CSS3 Fonts that are referenced share syntax with CSS2 and
are therefore very unlikely to change, and do not need to be prefixed.
Exact behavior of font matching may change for some atypical cases
when fallback is triggered.
4.4 CSS3 module: Lists
EPUB3 mentions this spec only for the semantics of some list style
types. The syntax of the list-style-type property is defined by CSS2,
so there is no need to prefix. As far as a list does not have so many
bullets (< 100), interoperability problems are insignificant.
Note: Reference to CSS3 List might be simply dropped, since the
current WD does not provide correct definitions.
4.5 CSS3 ruby module
EPUB3 use only one property in CSS3 ruby module. Its syntax will
change but the semantics are unlikely to change. No matter what we
do for other properties, it is probably wise to introduce a prefix for
this property, as agreed in the EPUB WG F2F in 2010-10.
4.6 CSS3 Speech
EPUB3 uses five properties, namely speech, cue, pause, rest, and
voice-family from CSS Speech. Daniel Weck wrote:
>'speak' has actually been divided into 2 separate properties, namely
>'speak' and 'speakability', but otherwise the functionality is the
>same and is unlikely to be disputed (only the name 'speakability' may
>change for a better one). Otherwise 'cue', 'pause, and 'rest' should
>be very stable indeed (they haven't changed since 2004).
5. Longevity
Longevity of EPUB3 publications is important for the EPUB WG. We
would like to ensure that EPUB3 publications containing properties
borrowed from CSS working drafts can be correctly rendered by future
EPUB reading systems.
However, it is probably too much to ask longevity for every possible EPUB3
publication. Most of the EPUB publications in the near future are
expected to use simple features of EPUB3 only. Longevity for simple
cases are crutial, while that for complicated cases are not.
It is also important to note that more than one approach may achieve
longevity. One approach is to adopt prefixed properties in EPUB3 and
make EPUB3 reading systems continue to support them. Another approach
is to rewrite EPUB publications by converting stylesheets when CSS WDs
become recommendations.
6. Forks
The CSS properties borrowed from CSS working drafts may become a EPUB
fork of CSS. Implementing both the upcoming recommendations and the
EPUB fork is a significant burden, especially when the differences are
not purely syntactical. It is also difficult for authors to learn both the
upcoming recommendations and the EPUB fork and use them correctly.
7. Possible options
(I'm still writing this part. Having received feedback, I think I can
send it today.)
--
Praying for the victims of the Japan Tohoku earthquake
Makoto