rel=canonical in specs

[Dominique Hazael-Massieux]

Hi,

A few months ago, I brought a patch to respec to make it document as a
link rel=canonical [1] the latest version of a given document:
https://github.com/w3c/respec/pull/938
(with the goal of having search engines return the "preferred" version
of a given spec)

The respec patch makes it configurable what URI to use as a latest
version, with a default to TR/foo - default that might need to be
revisited in light of Denis' proposals [2].

I'm looking for feedback on whether to push that approach more broadly,
which would entail first to add such a mechanism to other editing tools
(I would in particular be happy to look at providing a patch for
bikeshed), and maybe longer term make this part of the pubrules
requirements.

I have also separately started investigating how we could annotate our
huge set of existing TR documents with a rel=canonical - but it's likely
that if we come up with a system to do so, it would be more cost
effective for future documents to document it themselves.

Feedback would be appreciated. Thanks,

Dom

1. https://support.google.com/webmasters/answer/139066
2. https://w3c.github.io/tr-links/versioning/

[Tobie Langel]

Post by Dominique Hazael-Massieux
A few months ago, I brought a patch to respec to make it document as a
https://github.com/w3c/respec/pull/938
(with the goal of having search engines return the "preferred" version
of a given spec)

Thanks for doing this. It's an important step forward.

Post by Dominique Hazael-Massieux
The respec patch makes it configurable what URI to use as a latest
version, with a default to TR/foo - default that might need to be
revisited in light of Denis' proposals [2].

Indeed.

Post by Dominique Hazael-Massieux
I'm looking for feedback on whether to push that approach more broadly,
which would entail first to add such a mechanism to other editing tools
(I would in particular be happy to look at providing a patch for
bikeshed), and maybe longer term make this part of the pubrules
requirements.

+1 to both.

Post by Dominique Hazael-Massieux
I have also separately started investigating how we could annotate our
huge set of existing TR documents with a rel=canonical - but it's likely
that if we come up with a system to do so, it would be more cost
effective for future documents to document it themselves.

I'm still not convinced this wouldn't be better (and more consistently)
handled at the HTTP header layer for new specs, providing there's
agreement wrt to the discussion in [2].

Best,

--tobie

---
[2]: https://w3c.github.io/tr-links/versioning/

[Tab Atkins Jr.]

Post by Tobie Langel

Post by Dominique Hazael-Massieux
I'm looking for feedback on whether to push that approach more broadly,
which would entail first to add such a mechanism to other editing tools
(I would in particular be happy to look at providing a patch for
bikeshed), and maybe longer term make this part of the pubrules
requirements.

+1 to both.

Yup, same from me.

~TJ