The formal/stable WebP specification

[tricjury70]

It seems somewhat odd that the spec listed Google's developer page as a Normative reference for WebP.

Initially, this description was merged (#1347) with a warning about the instability of the spec, but the notes seems to have been removed (#1997) without much discussion since then.

Currently, the WebP specification is being developed by the IETF and is in AUTH48, the final stages of the publication queue.

While it will probably be a few more weeks before the WebP spec becomes Infomational RFC 9403, EPUB 3.3 is also in the final stages of its recommended track. Is there time left to revisit this issue?

[iherman]

@tricjury70 thanks for that note.

I think it is indeed too late for EPUB 3.3. The document is being voted on by the W3C membership, and the rule is that the text published as a Recommendation should be exactly what the membership voted for (except for possible spelling mistakes).

What I propose instead is to file this issue as an official Erratum. The current plan ahead is to form a maintenance Working Group that will take care of Errata and then republish a new version of the Recommendation. Although that WG will not be chartered to make any significant changes to the specification, issues like this one will clearly be in its purview. That also means that we can do the best thing and refer to the RFC when that one is published.

Would that work for you, @tricjury70 ?

cc @mattgarrish

[mattgarrish]

Ya, it would be nice to have a more formal reference for WebP, but I don't see any other option than a post-release fix.

[iherman]

I think this is a valid erratum, and should be marked as such.

I am not sure whether it is a class-1 or a class-2 change; caution would dictate to mark it as the latter.

@wareid @TzviyaSiegman @shiestyle @dauwhe ?

[mattgarrish]

Class 1 sounds like it's limited to markup changes that don't alter the text of the specification, but this is a respec shorthand leading to a new reference and bibliography entry. I expect it's class 2.

[mattgarrish]

Looks like this is still working its way through the RFC publishing process: https://www.rfc-editor.org/auth48/rfc9403

We'll have to wait longer.

[bduga]

Speaking with one of the authors for the RFC (J. Alakuijala), I have confirmed that the changes to the spec during transition to RFC are editorial in nature and should not require work by clients to handle any format changes. Apparently the last substantive change to the algorithm was a bug fix in 2014. Whether we want this to be class 2 or 3 is still an open question, but it seems like from a technical perspective, this is just updating the location of the spec, so it seems like class 2 is appropriate.

[iherman]

Thank you @bduga. This is important to note in case somebody questions our choice.

[mattgarrish]

Anyone know what happened to this? RFC 9403 appears to have been assigned to another standard.

The auth48 status page says the document should have been published, but searching the site for webp turns up nothing.

[mattgarrish]

Looks like it became an IETF draft: https://datatracker.ietf.org/doc/draft-zern-webp/

I guess we just leave this open and see if anything changes in the next six months. We can't cite it as is.

[mattgarrish]

Updating this, the IETF draft was published this month but it's yet another informational document. Looks like another 6 months to see if it becomes stable.