Open bansp opened 2 years ago
At the meeting today, we gave green light to this. Setting priority to try to handle this by the next meeting (no promises).
To be more precise, next to
<mimeType>application/xml</mimeType>
we would have something like
<mimeType status="unregistered">application/folia+xml</mimeType>
and then, instead of what should be:
<mimeType status="unregistered">application/tei+xml;format-variant=tei-iso-spoken</mimeType>
-- which is not fully precise, because the "application/tei+xml" is by all means registered -- we could do, with a twist, the following :
<mimeType value="application/tei+xml">
<mimeParam name="format-variant" value="tei-iso-spoken" status="unregistered"/>
</mimeType>
and we could (or should) push a step further, and do also:
<mimeType value="application/tei+xml">
<mimeParam name="format-variant" value="tei-iso-spoken" status="unregistered"/>
<mimeParam name="tokenized" value="{0,1}" status="unregistered"/>
</mimeType>
-- with a convention that {}
embrace a set of values.
A further step is to identify the registering entity (rather than just saying "yes" or "no"), and have the @registered
attribute as obligatory, and do:
<mimeType registered="IANA">application/xml</mimeType>
and then we would have something like
<mimeType registered="CLARIN">application/folia+xml</mimeType>
(it could also be @registered="no"
, of course, but see below)
Let's consider one further hypothetical step:
<mimeType value="application/tei+xml" registered="IANA">
<mimeParam name="format-variant" value="tei-iso-spoken" registered="CLARIN"/>
<mimeParam name="tokenized" value="{0,1}" registered="CLARIN"/>
</mimeType>
The final form
Meeting decision: we'll see if the CSC has anything to suggest here, and may at some point split this into individual tasks (processing in various ways -- at least two tasks: redefining the XPaths and providing a way to aggregate the parameters in some way; redoing the schema; checking what we can provide via the API and how).
Meeting decision: I'll hack at this in a separate branch.
Meeting decision: move to the next milestone.
I fully concur with the above decision: move to the next milestone.
Re-read, potentially open a new branch for this.
This is connected to #157
I am prioritising this one for myself, not to lose it from my sights.
This ticket is connected to #157 .
One stumbling block above was consultation with the SIC. I think the order should be different: first a prototype, then consultation. Otherwise the members will not be sure what they are being consulted about.
BUT the prototype has to come post-CAC, or in a separate branch.
If they are registered with some working pipeline architectures or recognized by the Switchboard, we want to catalogue media types and parameters, because they may be (or are) emerging best practices and then we