Re: respec drafts

On 17/07/2019 11:01, David Carlisle wrote:
> On 17/07/2019 08:08, Frédéric Wang wrote:
>> BTW regarding removing non-normative part, I was wondering whether it
>> makes sense to remove the RelaxNG schema from MathML Core:
>
> Yes I think it probably would be better to do that.
>
OK I guess I'll do that later if nobody complains.

>> That said I think it is still useful to have one (e.g. people using
>> XML tools or for the W3C validator) so maybe the RelaxNG schemas
>> should be moved and maintained in a separate git repository?
>
> Yes we have
>
> https://github.com/mathml-refresh/mathml-schema
>
>
> But it's currently configured to pull bits from the full and core spec
> but I could turn it round and make it self standing.
>
>
> Especially with things like onfoo and data-foo attributes you can't
> really specify them in general in a schema so having the schema
> normative doesn't really work, it more needs to be a framework in
> which you can make a schema for the instances you need. (I believe the
> w3c validator simply deletes any data-zzz attributes before passing to
> the schema validator for example)
>
Good point about data-foo attributes.

> A version of the core schema is already checked in to
>
> https://github.com/mathml-refresh/mathml-schema/tree/master/rnc
>
> so deleting the one from the core repository and editing that directly
> should be fine
>
If there is already another repo with schemas then we should definitely
not have one in the core repo.

I had done a few edits to better match the spec:

https://github.com/mathml-refresh/mathml-core/blob/master/mathml-core.rnc

and it is now out-of-sync with the version of the mathml-schema repo...

-- 
Frédéric Wang