[tei-council] Report on Sanity Checker + Exemplars

Lou Burnard lou.burnard at oucs.ox.ac.uk
Wed Jan 30 05:30:20 EST 2008


David Sewell wrote:
> Below is a report on running Sanity Checker on all the TEI P5 
> exemplar *.odd files in P5/Exemplars (plus tei_odds.odd, from Roma--
> where is that in the Subversion repository?).
>   
It is magically confected, I think.
> The "not reachable from root" warnings are all fixable, either by adding
> explicit <elementSpec mode="delete"> or perhaps by tweaking the modules.
>   
This is a really useful and helpful list! I suggest you should just go 
ahead and add the
required elementspecs: these schemaspecs were for the most part lashed 
together to test some particular constellation of modules and are not so 
lovingly hand crafted that they couldn't be improved on by almost anyone..

> For some reason, "handNote" and "teiCorpus" get defined in most of the
> generated schemas even when not usable (see #3 and 4 below). Can this
> be fixed in a global way, so they don't have to be explicitly excluded 
> from each affected exemplar?
>
>   
These are module-specific elements --  only available when a specific 
module is included. Until we revisit the topic of module dependency in 
ODD (may it be a long time), we just have to handle them specially I think.

> Under #5 below I list the "X does not exist" diagnostics that Sanity 
> Checker gives for various elements. I take it that these diagnostics are 
> intended to alert the user to customizations that have deleted elements 
> ordinarily used, so the Absolutely Bare schema generates lots of these, 
> all ignorable.
I may be wrong, but I think these are elements which are referenced 
explicitly by a content model rather than via a class declaration. We 
should probably get rid of them longterm, so I would rather leave the 
diagnostics in.

>  In other cases, for example with tei_odds and tei_math, 
> the diagnostics may be erroneous. It looks like Sanity Checker may not 
> be handing non-TEI-namespace definitions properly?
>
>   
That certainly needs more investigation

> I would suggest that as part of this week's cleanup we
>
>   * fix all the "not reachable from root" errors (except maybe for TEI Tite,
>     where we'll need input from Perry Trolard, see #2 below);
>
>   
I think he would say his work here is done...

>   * if possible, add some text to Sanity Checker preceding the lists of
>     "In X, Y does not exist" diagnostics to say that these are informational
>     only, not necessarily a sign of problems
>
>   
Or that these are warning messages only

> and then if Sanity Checker is in fact barfing on non-TEI-namespace elements,
> get that fixed as soon as someone who knows the code can do it.
>
>   
blatantly!

> David
> =======================================================
>
> SANITY CHECKER ISSUES with TEI Customization Exemplars
>
> 0. (Sebastian is looking at TEI Lite)
>
>   
In a separate message which I see I forgot to cc to anyone else, I've 
suggested that we should *not* make any changes to TEI Lite which would 
involve significant changes to the prose of its associated tutorial. The 
schema should not include elements which are not described in the prose, 
and if we don't have time to change the prose (which we don't) then we 
should change the schema accordingly. We can always do a new super 
improved TEI Lite as a distinct release later (though isn';t that what 
TEI Tite was supposed to be?)



> 1. TEI Absolutely Bare (tei_bare.odd)
>
>     Not reachable from root: refState, appInfo, application, handNote, geoDecl
>
>     Discussion: TEI Bare is intended to have an extremely simple teiHeader,
>                 consisting only of <titleStmt>/<title>, <publicationStmt>/<p>
>                 and <sourceDesc>/<p>. None of above elements are needed.
>   
I think you mean "IS needed"? :-)

>     Proposal:   delete them all using <elementSpec mode="delete">
>   
Yes: make it so.

> 2. TEI Tite (tei_tite.odd)
>
>     Not reachable from root: tagUsage, namespace, rendition, handNote
>
>     Discussion: for handNote, see section #3. Others are missing because
>                 <tagsDecl> is deleted from 'header' module.
>
>     Proposal:   If we decide to add <tagsDecl> back to TEI Lite,
>                 ask Perry Trolard's blessing for adding it here, too.
>
>   
No. See above.




More information about the tei-council mailing list