Repository navigation
xsd:anyURI #23
Description
Historically the W3C have had great difficulty in defining what a URL/URI was. the XSD 1.0 definition had some ... features, and XSD 1.1 redefined it to have the same value space as xsd:string (with the intention that you can use it with natural strings documents and any uri-encoding is done as some back end process cf a browser address bar.)
most of our existing cds have
<CDURL> http://www.openmath.org/cd/arith1.ocd </CDURL>
and the schema says
CDURL = element CDURL { xsd:anyURI }
so this matches or not depending on the built in anyURI definition.
Of the ones I have to hand rnv and xmllint says it is valid and emacs nxml mode says it is not
I don't really want to change the schema to make space explicitly allowed (doing that while referencing anyURI is a bit convoluted I think) and the schema is anyway in the openmath spec, so I suggest I just do a global edit and remove the space from xsd:anyURI elements. (I never liked those spaces anyway:-)
Activity
kohlhase commentedon Sep 30, 2017
I fully support this idea. I always hated the spaces.
kohlhase commentedon Sep 30, 2017
apropos nxml-mode, you have probably seen that I added the schemata under
lib/RelaxNGand addedschema.xmlfiles where needed.space arround URL issue #23
davidcarlisle commentedon Sep 30, 2017
sorry about that I'd got them locally in my checkout but not checked them in:-)
I think the reason I didn't check them in at the time was ...
We should formalise that the ones under the standard in the website repo are (like the cds) just copies from here, and I should get travis to copy them each time so we know we are in sync. which probably means copying back from the omstandard build some of the schema/dtd conversion logic.
Alternatively (perhaps better) we could consider the ones with the standard as authoritative and this travis could copy them from Openmath.github.io and check them in here if different..
or we could leave it as now and just sync by hand as needed....
kohlhase commentedon Sep 30, 2017
This sounds like the answer to the other issue: #24
But just to be clear:
This is contrary to my assumptions (and I think what is documented) I always had the idea that
I think we should become very clear about the mode and document that, so that things are transparent and I do not forget.
davidcarlisle commentedon Sep 30, 2017
yes content dictionaries are copied over from CDs repo to Openmath.github.io rep and checked in.
So the ones in CDs are the sources to edit although arguably the ones on the openmath,org website are authoritative (as that's the URL given by
<CDURL>) but the only ones that should be edited are in CDs.however the standard document itself has master version in the OMSTD repo and currently the schema are built as part of the standard.
Currently the schema are synced between the two repo by hand.
It would be possible for travis to do this. (either the OMSTD travis or the one here)
Separately we could consider schema development to be part of CD development and so edit them here and push to the website rather than pull from OMSTD. I half suggested that above but on balance I think it is correct that the schema is tied closely to the standard.
kohlhase commentedon Sep 30, 2017
Ah, you were talking about the schemata, now it makes sense to me.
I agree that schema development should be part of the standard development and not of the CD development. I also think by-hand-sync for the schemata is better.
Let us document this in the READMEs of the various repositories. I will make a first draft and then you can interate?
davidcarlisle commentedon Sep 30, 2017
OK