Re: Minutes 2018-11-26

> On 28 Nov 2018, at 01:44, MURATA Makoto < <>> wrote:
> Folks,
> I am sorry for not attending the teleconf this week.
> Tzivya and Dave appear to believe that the testing process
> for the REC track will provide more interoperability.  I
> am not sure if I agree.
> At best, the testing process provides implementability of
> the EPUB 3.2 specification.  It does not necessarily lead
> to interoperability.
> First, are we going to test HTML, CSS, and SVG features
> as part of the EPUB 3.2 REC track? 

It is not our job to test other standards from a process point of view.  

We may (should?) have an HTML validator in EPUBCheck, of course, and we may (should?) try to see whether the Web Platform tests can be used for reading systems, too, but that is a different matter.

> Or, are we going to
> test EPUB-specific features only?  I know that more than
> a half of the interoperability problems in Japan are around
> HTML and CSS. 
> Second, are we going to pick two particular implemenations
> for each feature?  Or, are we going to test if an 
> implementation supports all features?  

Although the details are to be specified by us, basically no. 

Maybe it is cleaner if, as an example, we look at the implementation report of the annotation model specification: <>

12 implementations were considered (see the big table down the page). The group separated required and optional features and a number of corresponding tests. The only requirement to pass the Rec was that the required features were implemented by at least two implementations. Although, in this case, there were implementations passing all the required tests, this was not a requirement, see the explanation on the exit criteria. Note that the report content (ie, the columns) were provided by implementers themselves.

> If we pick two
> per each feature, will our test have any impacts on
> interoperability?

Again, let us separate what the W3C process requires and what the group may choose to do extra.

Formally, the W3C process proves that all the required specification features can be implemented in a vendor independent way. That is it.

However, the testing can be as rich as we want beyond that, as demanding as we want: it is a good goal to ensure interoperability by having implementations that have a full feature set implemented. That is entirely dependent on the group's choices.

I hope this helps


> Regards,
> Makoto
> 2018年11月27日(火) 1:21 Ivan Herman < <>>:
> Here they are:
> <>
> Cheers
> Ivan
> ----
> Ivan Herman, W3C 
> Publishing@W3C Technical Lead
> Home: <>
> mobile: +31-641044153
> ORCID ID: <>
> -- 
> Praying for the victims of the Japan Tohoku earthquake
> Makoto

Ivan Herman, W3C 
Publishing@W3C Technical Lead
Home: <>
mobile: +31-641044153

Received on Wednesday, 28 November 2018 05:51:25 UTC