- From: MURATA Makoto <eb2m-mrt@asahi-net.or.jp>
- Date: Sat, 29 Jul 2017 04:55:51 +0900
- To: Romain <rdeltour@gmail.com>
- Cc: W3C Publishing Working Group <public-publ-wg@w3.org>
- Message-ID: <CALvn5EDy=q1wue_A+nfApWPc2F6kVpKKo-fXvGRJUT9GnP0ERg@mail.gmail.com>
>- assuming that the manifest is an external non-HTML document (read JSON), its URL is different from any of the URLs to the HTML content. I do not agree. One option for representing manifests is to augment navigation documents, which is HTML. Regards, Makoto 2017-07-28 1:31 GMT+09:00 Romain <rdeltour@gmail.com>: > In another thread Hadrien said: > > > "[the] URL [of the manifest] identifies the publication as a whole", > that's where we have a pretty massive disagreement as I believe that the > URL of the manifest is a perfect fit to identify the publication as a > whole, but others (Dave, Garth for example) want a URL that returns HTML > instead (not sure why an identifier MUST return HTML, but anyway...) > > > I think I was in the "others" group, but I believe there's a > misunderstanding about what we're talking about. > > What we all probably agree on the obvious: > > - a publication has a manifest, which is addressable (it has a URL) > - a publication has HTML content ("primary resources", it seems), which is > addressable (each primary resource has a URL) > - assuming that the manifest is an external non-HTML document (read JSON), > its URL is different from any of the URLs to the HTML content. > > > I can see how a UA will want to directly process URLs to non-HTML > manifests, for instance to parse the catalog from a store. > I would also hope that a web site pointing to a Publication would use a > URL to an HTML resource (so that the link can be opened and rendered in a > non-supporting browser). > > But these are use cases for **links**, not **identifiers**. > > So, my question is: what is the envisioned use case for an "addressable > identifier"? does it matter or can we assume that there's no need for a > canonical location and we just point to whatever makes the most sense > depending on the context? > > Romain. > > PS: the PWP UCR contains one use case about uniquely identifying a Web > Publication [1], but I'm not sure I see to which extent this use case > requires more than the ability to link to the content. The UC doesn't make > a strong case for the uniqueness requirement. > > [1] https://www.w3.org/TR/pwp-ucr/#unique-identifier > > > > -- Praying for the victims of the Japan Tohoku earthquake Makoto
Received on Friday, 28 July 2017 19:56:15 UTC