clarin-eric / standards

work space for the Standards and Interoperability Committee
https://www.clarin.eu/content/standards
3 stars 13 forks source link

multiple <domain>s inside a single format -- should we think of handling that? #66

Closed bansp closed 1 year ago

bansp commented 2 years ago

Debugging some files, I came across stuff like

        <format>
            <name id="fComa">EXMARaLDA Coma</name>
            <domain>Metadata</domain>
            <domain>Contextual Information</domain>
            <level>recommended</level>
        </format>

which is a way to be concise and definitely editor-friendly, but which kills the display because rf:print-centre-recommendation() in recommended-formats.xql doesn't handle multiple domains.

I think, for now, we should probably keep the format-recommendation format simple, with a single domain per format, partly because otherwise it may be hard to say what the <comment> refers to (this very statement could be argued with...).

But maybe it's something to consider, later on?

bansp commented 2 years ago

One way in which this could be considered is as a shorthand, provided that it's followed by applying the sorting script (just recently added by Eliza). Assuming that the encoder does something like (and I'm already assuming that we've moved away from using <name>):

        <format id="fComa">
            <domain>Metadata</domain>
            <domain>Contextual Information</domain>
            <level>recommended</level>
            <comment>[something true of each domain separately]</comment>
        </format>

The sorting script could (however that's implemented):

In the case above, the result would then be:

        <format id="fComa">
            <domain>Contextual Information</domain>
            <level>recommended</level>
            <comment>[something true of each domain separately]</comment>
        </format>
(...)
        <format id="fComa">
            <domain>Metadata</domain>
            <level>recommended</level>
            <comment>[something true of each domain separately]</comment>
        </format>

That could surely be sold as a kind of shorthand for those entering the data manually.

bansp commented 2 years ago

Eliza and I decided to mark this as a "won'tfix" for now, due to the many potential complications that such a move would open. Part of the issue is that it would be tempting, in the interest of time, to go for the most convenient denominator, which might mean the most general format such as "XML" or "TEI", which in turn is meaningless.

bansp commented 1 year ago

Closing as wontfix.