[MPEG-OTSPEC] Updates to specification

[Norbert Lindenberg]

I agree with John that it would be useful to look at how the groups working on some other essential standards decide what goes into their specifications, and when. John, can you point us to a reference for the W3C independent implementations requirement?

Here’s the process document of ECMA TC39, the group in charge of ECMAScript, the language better known as JavaScript:
https://tc39.es/process-document/

Note:

– There are clearly defined stages from idea to standard, with criteria for each stage.

– The criteria relate not only to the spec itself, but also to the conformance test suite Test262, to compatible implementations, and even to feedback on those implementations.

The Unicode Consortium doesn’t seem to have a similar written document, but it produces the Internationalization Classes for Unicode, their reference implementation, which Google, Apple, IBM, and pretty much everybody else relies on, and which usually ships within weeks of a new Unicode version.

Best regards,
Norbert
Lindenberg Software LLC



> On Aug 13, 2020, at 19:21, Levantovsky, Vladimir <Vladimir.Levantovsky at monotype.com> wrote:
> 
> On Thursday, August 13, 2020 6:08 PM John Hudson wrote:

>> Vlad, does ISO/MPEG have any kind of implementation requirement similar to, say, the W3C independent implementations requirement? Is it possible to institute one?
>> 
>> One of the things that has concerned me about both the OT and OFF is that they have sometimes been what I would class as speculative specifications, and have numerous features that are formally part of the format but have never been implemented (Peter just reminded us of one). In recent years, there has been some improvement in this regard, e.g. the implementation many aspects of OT variations before the 1.8 spec was published, but so far as I am aware there is no requirement for implementation built into the processes, and nothing like a public draft stage such as CSS has.
> 
> Well, while there are no formal ISO requirements related to implementations of the spec, we can institute some. MPEG has always been in favor of complementing the spec with reference implementations so that an implementer has something else to look at if spec language isn’t 100% clear. But we also need to be mindful about the effects of these requirements – speaking about W3C work, the WOFF2.0 spec development was finalized in 2014, however, the Recommendation has been ratified only in March 2018. The delay in final ratification was only due to work related to implementations and conformance testing. While it didn’t really hurt WOFF2 adoption, and many browser vendors don’t mind implementing the technology specified as part of the Working Draft, ISO process is different.
> 
> Also, some features may be so specific to particular scripts / layout needs that the implementation requirements would be hard to impose or enforce. Or, a feature that is optional in nature that may be useful and nice to have but the one that is never been used – do we really want to get it out of the spec? We can definitely deprecate things that are not needed / useful.
> 
> Thanks,
> 
> Vlad