Re: consuming the semantic tag (was Re: Subscripts in Content MathML)

Stan,

there is definitely an "Accept" http-header which all browser send and 
specifies the list of mime-types a browser accepts. This header does 
indicate, for example, whether a picture format is supported... Why not 
enrich the declaration of the support for mathml-content or openmath 
(with their respective mime-types) with "mime-type-attributes" which 
specify the set of supported symbols (or content-dictionaries) ?

Making it a language would require an extra http-header, or do I 
mistake ?

Now, there's an extra issue here: would a browser declare that it 
supports the mathml-content-of-maple ?? How would it know it ??

paul



Le 13 oct. 05, à 03:47, Stan Devitt a écrit :
> The mime-type is really just a passive server-side message identifying
> the content provided by the server as having certain characteristics.
> The client then can decide how it is going to handle it.
>
> Multi-language support works the other way round.  The client 
> advertises
> its self as being (say) french, and the server decides how to respond.
> (look at a typical apache server httpd.conf file)
>
> If the client could adverstise itsself as supporing (say)
>  "mathml-content:maple" then the server could strip out those
> semantic fields that do not match.
>
> It is the "language" model that would allow us to have one master on
> the server but serve up filtered versions of the document to our PDA.
>
> The main difference is that it will be common for the client to
> want to support multiple languages (at least presentation and
> content, but possibly more as in presentation, content, maple, 
> mathematica
>
> Stan.
>
> On Thu, Oct 13, 2005 at 12:20:32AM +0200, Paul Libbrecht wrote:
>>
>> I agree the mime-type of the content of a semantics element is to be
>> disjoint of the mime-type of the document but don't you think that the
>> type of semantics element (call it a label, an annotation, or...) is
>> close to a mime-type and further than a language ?
>>
>> A mime-type describes possible software operation whereas a language
>> can only be made acceptable or not by a human.
>> I think it is fair for a computer algebra system to describe through
>> mime-types that it suports a set of symbols...
>
> ...  and that is no different that a browser saying it supports french.
>
> client X supports vocabulary Y.
>

Received on Thursday, 13 October 2005 08:55:08 UTC