Re: Updated WOFF2 spec and CTS planning

Hello Vlad,

[I'm just back from vacation today]

Monday, August 11, 2014, 7:04:17 PM, you wrote:


> Please see attached the updated spec for your review (sorry, my CVS
> client acting up and I can’t get through to upload the spec).

We really need to get those CVS access issues sorted out. Let's
discuss offlist, bring in systeam again if there is any change from
last time, and if all else fails I can show you the method I use on
Windows (command line CVS, in a cygwin shell).

The version you mailed in was based on the revision 1.6 CVS version.
There were changes since then (latest revision was 1.17). Luckily most
of the subsequent changes were localized to the header material (style
sheet and status changes) and to an appendix (MIME type) and
references (updates) plus some small markup changes (HTML5 does not
have the tt element).

To merge these, I started with Vlad's mailed-in copy (because the
conformance identification changes are throughout the body of the
text), re-added the MIME appendix, merged the references sections
(Vlad's had CSS3 Fonts, cvs had an update to -01 versiun of Brotli
Internet draft) and re-added the status section. I merged in the style
changes from cvs with the ones added to do conformance styling in
Vlad's copy.

I then checked this into CVS
http://dev.w3.org/webfonts/WOFF2/spec/
W3C Editors Draft 18 August 2014

As far as I can see there were no changes to the normative body of the
spec since the version on which Vlad's edits were made. I'm continuing
to check with diff in case there were, in which case I report the,m on
this list and will backport them to CVS.

CVS log to date is attached.

> The spec is marked up to identify the conformance requirements,

Its the same markup and styling as we used for WOFF 1.0, for those who
were part of that effort. It means that each conformable statement has
an ID so it can be linked to directly. In addition, there is a class
indicating whether the conformance id on Authoring Tools, User Agents
9browsers) or the File Format itself. This affects how it is tested.

The styling adds a marker before each statement, and on hover
highlighting which is also triggered if the statement is linked to
directly.

> and
> after going through this exercise I am afraid that we are severely
> “underspecified” when it comes to UA and authoring tool requirements
> and testable assertions.

I agree. The benefit of marking and styling them is that we have a
better idea of what needs to be added (or more commonly, reworded so
that it becomes testable).

>  There is  still a lot of things to do on
> the spec side before we can even remotely consider it ready for prime time.

Right. Note that these changes typically don't affect running code;
they just make explicit the requirements so they can be tested and so
we can prove that multiple implementations are interoperable.

> Note to Raph – if you can find some time and go through the spec
> with your additional edits and changes, I’d really appreciate it.

Raph, I'm assuuming you will use CVS to do that.

> There are still grey areas regarding “nominal size” definition and
> decoder behavior regarding memory handling  due to potential size
> variations of decompressed and reconstructed font data.
>
> All, I’d also very much appreciate your input on what parts of the
> spec we can tighten up to add new testable assertions and UA /
> decoder requirements for CTS – once we complete this first step we
> should be able to start the CTS planning  and decide what tests need
> to be in place and how we can / should test each condition.

Yes. I'm starting on that today, but wanted to send this mail first so
we are all working off the same copy
http://dev.w3.org/webfonts/WOFF2/spec/
W3C Editors Draft 18 August 2014


> I would like to get this discussion started now, before we meet for
> our F2F on Sep. 16. Email your comments to the list, and let’s
> discuss them during the telcon this week. FYI- I will be on vacation
> (and out of reach) during the next week  so the next opportunity
> (other than this week, i.e.) to discuss this would be Aug 27.

Okay.


-- 
Best regards,
 Chris                            mailto:chris@w3.org