– DRAFT –
Fonts WG

<chris> I raised 5 issues. 3 should be addressed before publication. I have simple PR for all of them. The other two are just so i don't forget

<chris> https://github.com/w3c/IFT/issues

<Vlad> scribenick jpamental

Garret: presenting the spec, found here https://w3c.github.io/IFT/Overview.html (covers patch-subset, range request is coming)

<chris> It also covers the protocol negotiation

Garret: Myles will be contributing range request

Garret: pointing out we're using CBOR for encoding

Garret: pointing out SparseBitSet section along with IntegerList data structures

Garret: showing specs for subsetting variable axes

Garret: Sections on Client and Server to describe desired behaviors on both parts, but intentionally broad, without getting overly specific on implementation

Garret: server responds to request from client by sending more than the bare minimum of the client request, but implementor can decide on how much more in order to be more efficient with subsequent requests

Garret: Showing miscellaneous sections at the end covering computing Checksums, Codepoint reordering, and patch formats

Garret: Mentioning how to implement in HTML/CSS to let the author suggest how the transfer should work in the font-face declaration

<chris> https://github.com/w3c/IFT/issues

Chris: mentioned that he had raised 5 issues on the spec, at least 3 of which should be resolved before publication

Myles: I also have some questions

Chris: Proposing some language changes under Privacy and Security considerations

Garret: These look good, I will review the PR

Chris: link cleanup, adding the w3c ID

Garret: I will review that PR

Chris: The section titled 'In HTML', but it's not accurate. Split into 2 sections 'One in CSS', another about protocol negotiation

Garret: agreed, will review the PR

Myles: You mentioned 2 different hash algorithms. what are they?

Garret: These are just checks to make sure things are working correctly

Myles: what happens if they don't match?

Garret: The client is supposed to re-request from scratch if things don't match up. Expected that this will be rare, but better than allowing errors to accumulate

<chris> Ok so they re-request to get a known good font, seems correct to me

Myles: I'm worried about multiple back-end servers handling subsequent requests (like at Google Fonts). Maybe there should be a way to ensure one server (with the same software and files) get fulfilled with same versions, etc

Garret: We have a 'sort of similar' idea in Google Fonts, but not quite. I'll think about that

Garret: I'm not certain this will be an issue, but will look into it further

Myles: If the second server has different versions of software it might not be able to produce a new patch correctly

<chris> we should raise this one as an issue

Garret: You may be right, I'll follow up and detail a new issue on GitHub

Myles: This doesn't need to stop us from publishing now, but we just have to follow up

Vlad: are there any objections to publishing this as the first working draft?

Vlad: Don't forget that this triggers the first step in patent review

<chris> Also if there are no patent concerns, there is nothing that needs to be done

Vlad: I hear no objections

Resolution: Publish FPWD of Incremental Font Transfer

<chris> so actually, https://github.com/w3c/transitions/issues/374

<chris> (we explain W3C Patent policy)

<chris> (we explain W3C Wide Review)

<chris> (we explain test driven development)

<chris> Vlad: TPAC happening

<chris> https://www.w3.org/2021/10/TPAC/Overview.html

<chris> vlad: opportunity for demos. Can we do one?

<chris> Garret: we could do an in-browser demo to gain awareness

<chris> ... deck used at ATypI technical talks

<chris> (adjourned for realz)