application/mathml+xml correct in media types?

w3c / mathml-docs

Notes and other documents by the Math WG. See also https://github.com/w3c/mathml-core (repo for the MathML Core spec) and https://github.com/w3c/mathml (repo for the MathML 4 spec).

https://www.w3.org/Math/

Other

4 stars 10 forks source link

application/mathml+xml correct in media types? #38

Closed NSoiffer closed 2 years ago

NSoiffer commented 2 years ago

I'm not up to date on media types, so this may be a stupid question...

In https://w3c.github.io/mathml-docs/mathml-media-types/#media-type-choosing, the media types are all application/mathml... +xml. Is +html allowed? If so, seems like that should be mentioned.

NSoiffer commented 2 years ago

I just clicked through to RFC3023 and see that only has Application/mathml+xml. This doc is 2001. Is there a better/more up-to-date reference?

davidcarlisle commented 2 years ago

3023 is just the general foo+xml scheme, it doesn't get updated for specific instances of xml media types. The currently registered up to date reference for the 3 mathml types is the appendix in MathML3 that this would replace. There is no corresponding +html scheme, you'd use text/html to get html parsed MathML

polx commented 2 years ago

Right. The +xml suffix is defined by RFC 3023 (and indeed MathML's generic content-type as well). RFC3023 is important when registering a +xml type because it defines a few aspects of the transmission (including the somewhat difficult charset family of parameters).

The current list of all registered media types is displayed at the IANA: https://www.iana.org/assignments/media-types/media-types.xhtml It is regularly updated.

That list would be updated with a different URL once we publish an independent recommendation with the MathML-media-types.

NSoiffer commented 2 years ago

Thanks for the corrections -- I guess I already knew that...

One of my concerns with RFC3023 is that it says

MathML documents are XML documents whose content describes mathematical information, as defined by [MathML]. As a format based on XML, MathML documents SHOULD use the '+xml' suffix convention in their MIME content-type identifier. However, no content type has yet been registered for MathML and so this media type should not be used until such registration has been completed.

I think it would be a good idea to include the IANA ref/link so that people can easily find that the registration has been completed.

davidcarlisle commented 2 years ago

@NSoiffer hmm RFC can't be changed, only errata or a new one obsoleting the existing one.

Turns out 3023 has both

https://www.rfc-editor.org/errata_search.php?rfc=3023

says 3023 is obsoleted by 7303 that removes mention of mathml as an example +xml type but it has status "proposed" although dated back in 2014. Wikipedia seems to take it as authoritative though https://en.wikipedia.org/wiki/XML_and_MIME perhaps we should replace the 3023 reference with 7303, @polx ?

NSoiffer commented 2 years ago

This seems like a mess: the old document is official but obsolete and the new document is "proposed" and doesn't mention MathML.

I'm not up-to-date on that part of the standards world. Is it reasonable to just mention 3023 since it includes the mentions of some transmission details and immediately after one of the refs say something like: "The MathML media types have been registered with IANA" and give the link in the bibliography?

polx commented 2 years ago

We shouldn't expect an RFC to mention MathML. It will be outdated at some point. The list is the source to mention MathML and to be updated.

As for the evolution of 3023, this is discussed currently, but with respect to the possibility of registering new suffix types (json was suggested, several suffixes were discussed). Do we have a reason to try to bring back MathML? I think rather no.

davidcarlisle commented 2 years ago

@NSoiffer, as @polx says +xml RFC not mentioning MathML is good I think. It was only mentioned in 3023 as a speculative example as a there were no actual +xml vocabularies at the time, and as shown here having examples in the generic rfc that can't be changed just causes issues down the line.

polx commented 2 years ago

It was discussed today and decided:

to add a reference to the revisions of RFC3023 everytime we reference RFC3023 (at least the latest)
to make the "list of media types" a reference as well (maybe with a "last checked" comment)

polx commented 2 years ago

I have now:

thinned the set of unused references in the header
reference the registration process and the list of media-types as bib entries
converted references from rfc3023 to rfc7303 (which obsoletes it)

I believe that this addresses all issues mentioned here.