Initial draft of the Open Screen Application Protocol.

[markafoltz]

Initial draft of the application side of the spec, following plan in #341.

[backkem]

This is a clean initial split.

It may be worth considering if part of the agent-info concerns are better handled on the 'network' side of the split. E.g.: it's hard to know what you're connecting to/authenticating with without giving it a proper name. But this can be refined in iterations.

[tidoust]

I just spotted a couple of metadata updates to do.
I haven't reviewed the entire application.bs file but that seems a good start :)

[markafoltz]

It may be worth considering if part of the agent-info concerns are better handled on the 'network' side of the split. E.g.: it's hard to know what you're connecting to/authenticating with without giving it a proper name. But this can be refined in iterations.

@backkem I noted this while working on the network side of the spec and filed #346, will post your comment there.

[anssiko]

Thank you @markafoltz for getting the split drafts out ahead the meeting.

I'd be supportive of merging this and #347 with an Editor's Draft status to facilitate review. The group decision time is when we want to publish to TR. The local builds are clean, but it'd be convenient to be able to hand out URLs to rendered documents to reviewers outside the group as well, and deploy would enable that.

@tidoust build/validate/deploy for multiple specs since spec-prod v2.3.0 do work and the workflows look good to me, thanks for tweaking those. As a secondary non-blocking consideration: PR Preview does not yet support repos with multiple specs tobie/pr-preview#18 -- are you aware of workarounds (besides separate repos)? Maybe we can live without PR Preview for a while...

[anssiko]

@tidoust application.html deploy to gh-pages was skipped resulting in https://w3c.github.io/openscreenprotocol/application.html

Build and validate passed.

Looking at https://github.com/w3c/spec-prod/blob/main/docs/examples.md#multiple-specs-in-same-repository I believe a possible fix is to add this line:
DESTINATION: application.html

After this line:

openscreenprotocol/.github/workflows/auto-publish.yml

Line 57 in 659582c

SOURCE: application.bs

(And similarly for the network part.)

[tidoust]

I believe a possible fix is to add this line: DESTINATION: application.html

That may work but, if it does, I don't understand why ;)
DESTINATION should take the value application.html all by itself when SOURCE is application.bs.

I'm looking into it but don't know yet why the job skipped deployment to gh-pages.

[tidoust]

I'm looking into it but don't know yet why the job skipped deployment to gh-pages.

Actually, that's easy, we're just not looking at the right logs ;) These logs are for the PR itself, where the spec-prod action validates things but indeed does not deploy anything (on purpose, that's a PR). The right logs to look show a Bikeshed validation error because nothing build application_messages.html. That's what needs fixing. I'll prepare a PR.

[tidoust]

The right logs to look show a Bikeshed validation error because nothing build application_messages.html

More accurately, the file gets generated, but the action resets the status of the git folder after taking care of the first document (through a call to git clean), which deletes the other generated HTML files and makes the following generation of application.html fail. The action works best in completely separate jobs. Anyway, I'll propose a rewrite.

[tidoust]

Publication now fixed. Both drafts are now available at:
https://w3c.github.io/openscreenprotocol/application.html
https://w3c.github.io/openscreenprotocol/network.html