[sword-devel] TEI markup support

[DM Smith]

On May 12, 2008, at 7:10 PM, Chris Little wrote:

> Our plans are to use TEI for dictionary encoding from here forth.

Chris has started a wiki for how to encode a TEI dictionary. See: http://www.crosswire.org/wiki/index.php/TEI_Dictionaries

In Sword's svn you can find tei2mod to build the module. It is  
discussed on the wiki page.


> At the
> moment we have some support for both P4 and P5 conversion to RTF (used
> by BibleCS), plain, and HTMLHREF (used by GnomeSword, BPBible?,
> others?). So...
>
>
> Issue 1:
>
> Which filters remain necessary before we can declare that we support  
> TEI
> and ship 1.5.11 with TEI support sufficient for all of the major  
> frontends?

It would be nice to get the <ref> element working at least for Bible  
references and if time permits, internal references.

We have code that models this already.

>
>
> I assume no one parses GBF, ThML, or OSIS directly for rendering. Does
> anyone use the plain HTML filter? (I'll tackle the WEBIF filters and  
> do
> any revisions to the HTMLHREF that seem necessary.)
>
>
>
> Issue 2:
>
> I think DM and I (so far CrossWire's only 2 TEI encoders) are agreed  
> on
> using the more recent TEI P5 for CrossWire-encoded texts. None of  
> these
> are yet available publicly, but DM's NASB lexicons use P5 now and my
> (coming soon) revision of Webster's Dictionary uses it.

Yes, we are agreed. :)

>
>
> Everything currently posted uses P4 (which was current at the time  
> they
> were encoded). That includes stuff from Perseus (which came to us as
> TEI) and things from the Germanic Lexicons Project (which were encoded
> in TEI by me).

These are all beta. And no released frontends support TEI, so any  
decision has little impact.

>
>
> Should we:
> a) support TEI P4 and P5 separately (so we would need TEI P4 and P5
> flavors of the filters targeting RTF, HTMLHREF, WEBIF, plain,
> etc.)--This would require extra work and a larger memory footprint  
> than
> the other options.

No.

>
>
> b) support TEI P4 and P5 jointly (one filter for TEI, irrespective of
> version, for each target markup)--This would be possible because  
> there's
> not that much significant difference, but would be slighly wasteful.


In looking at the TEI filters that we have today, I think that  
everything in there is also P5. The reason I favor this is the example  
(I don't remember where it is) at crosswire.org of remote fetching of  
entries from another website and displaying them. I think it used our  
filters. If this still makes sense, then P4 is likely to be a target  
too.


>
>
> c) convert TEI P4 docs to P5
>
> I think I prefer option c. It shouldn't be that difficult given the
> standards' similarity.

I find them very similar. I think that we should settle on one style  
of cross-references as you and I discussed earlier.

You might find that validating them with your P5 schema will pass.

In Him,
	DM