Comment by gaxun | Hacker News Reader

Comment by gaxun | Hacker News Reader

I spent about 40 hours in December and January implementing a browser extension for Chrome and a server that speak the web annotation protocol and use the web annotation data model in this specification.

It was very easy to pass the tests the W3C working group used to verify that they had two working implementations of the data model and protocol. Most of the test default to passing if the specified tag is not present. Basically, it's not clear whether a serious, real attempt to use this has been made. I'm unconvinced that the specification is robust enough to be useful without ending up with a lot of vendor lock-in.

The toy extension was playing around with using these annotations to alert publishers and potentially other users of typos in their articles and pages. It would be nice to have a side channel to report typos other than just using the comment section or trying to find an email address. Will the "meta web" ever catch on?

I never published it but I still might add a page about my experience on my website. I have posted about the idea there before.

I'm not sure where else to complain so I will just leave this example here. I was frustrated reading the specification because it contradicts itself.

The data model has a required field called `id` which is an IRI (like a URI) that is basically a globally unique identifier.

The protocol allows an annotation to be transmitted without the `id` field attached.

Why? Is the field required or not.

In my toy implementation I had my browser client attach a v4 UUID as the `id` field before sending it to my server. But it would have still been valid without it.

> I'm not sure where else to complain so I will just leave this example here.

All of the specs, in their "Status of the Document" section, say:

> This document was published by the Web Annotation Working Group as a Recommendation. If you wish to make comments regarding this document, please send them to public-annotation@w3.org (subscribe, archives). All comments are welcome.

I'd guess, therefore, you should complain there (I have no idea if you have; I haven't looked through the archives!). Of course, there are plenty of W3C groups where specs have became pretty much totally abandoned as soon as they've reached REC, so it's totally plausible nothing will happen. :( (This tends to come about because groups are chartered to work on specs and bring them along the REC-track till they reach REC; unless a further version is being worked on there isn't necessarily any group actually with maintenance of the spec in-scope.)

I tried emailing two of the names listed at the time, for two different purposes. I got a little bit of a response from one, nothing from the other.

I read the "rules" for W3C groups and in order to actually participate you need to be a member of a big organization and all this other stuff. I'm not sure anyone would have listened.

Almost all W3C groups nowadays do almost all their work in public on public mailing lists; actually being a member of the WG is rarely a requirement for participation.

And if it was sent between the spec becoming a Candidate Recommendation and going to Proposed Recommendation, it must (in theory) have been addressed. If not, something's gone wrong process-wise with how the group was operating (and from poking around a bit, it seems likely it did). Le sigh. :\

Thank you for the explanation. This has been my only experience with the W3C system and as an outsider it was very intimidating. I'm certainly willing to accept the idea that I wasn't going about things the right way or understanding what I was reading!

Many of the smaller, newer groups with fewer people with a background in the W3C end up being somewhat dysfunctional and with odd processes and that almost certainly makes it feel harder to participate than it should be.

From prodding around a bit (notably [0], which sadly is in Member-only space, but plenty of administrivia is there, and in principle no technical work for almost all groups), it seems like every issue reported to the Working Group (regardless of where) should have ended up with a GitHub issue, with [1] being meant to have been all issues while the specs were in CR.

Pointing in the specs to a mailing list to report issues, and then relying on someone to copy them into GitHub, seems doomed to fail: it's far, far too easy for one thing to not get copied. Really the "Status of the Document" should've pointed to GitHub for new issues being filed (possibly with a fallback to the mailing list for those unable to use GitHub for organisational or other reasons).

[0]: https://lists.w3.org/Archives/Member/chairs/2016OctDec/0143.... [1]: https://github.com/w3c/web-annotation/milestone/3?closed=1

I've had this exact same urge for years now and have had a couple of false starts at tackling the issue. Glad to see 1) I wasn't the only one wanting to make those simple copy edits and 2) someone followed it through and did something about it.

Reply on news.ycombinator.com