[tei-council] Constraints on anyURI

[Sebastian Rahtz]

On 19 Nov 2012, at 22:51, Martin Holmes <mholmes at uvic.ca>
 wrote:
> I think Saxon is probably playing it safe. As the spec says, there's no 
> way to reliably resolve a URI, especially given the range of registered 
> and unregistered URI schemes that exist, so the best we can probably do 
> is check for some things we know must be wrong (such as spaces, commas 
> and perhaps percent signs which don't precede hex numbers). If we do 
> this through Schematron it will presumably work for everyone that's 
> using RNG with embedded Schematron.
> 



trouble is, when I tried this,. it was clear that Saxon is doing url-encoding
for me. the output of resolve-uri() from

n-/////CTL
foo:n-CT::#:L
http:/n-CTL t-TR Ø-OBJ, n-SUBJ

was 

file:/Users/rahtz/TEI/Sourceforge/trunk/P5/Exemplars/n-/CTL
foo:n-CT::#:L
http:/n-CTL%20t-TR%20?-OBJ,%20n-SUBJ

So it is hard to catch it out!
--
Sebastian Rahtz      
Director (Research Support) of Academic IT Services 
University of Oxford IT Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431