Re: [apps-discuss] draft-pbryan-zyp-json-pointer: name syntax for non-ASCII

I don't think this will work because every JSON parser I've used so far
processes "\/" as solidus "/". By the time it's parsed, there is no
escaping left. Or did I misinterpret, and in actual fact the backslash
would be doubled in actual JSON expression (e.g. "/Bjørn\\/Carsten\\/foo
bar")? That could work. 

Paul

On Thu, 2011-11-24 at 16:42 +0900, "Martin J. Dürst" wrote:

> On 2011/11/24 10:59, Martin Thomson wrote:
> > On 24 November 2011 08:37, Carsten Bormann<cabo@tzi.org>  wrote:
> >> As usual, the guiding principle is that easy things should be easy, and that hard things should be possible.
> >> Complicating the usual case (the only one that many people seem to have thought about yet) is indeed a big no-no.
> >>
> >> Grüße, Carsten
> >
> > A reasonable suggestion.  Maybe just drop the quotes and force an
> > escape for / instead of ":
> >
> >    /Bjørn\/Carsten\/foo \uD834\uDD1E
> 
> For the moment, cutting off the surrogate pair at the end, and adding 
> something to keep the space visible:
> 
>      /Bjørn\/Carsten\/foo bar
> 
> 
> Yes indeed. So when defining json pointers, use '\/' to escape a '/' 
> (and use '\\' to escape '\') that's part of a string rather than a 
> hierarchy delimiter:
> 
>      /Bjørn\/Carsten\/foo bar
> 
> Then when using that as a fragment identifier, you have to escape the 
> '\' (not allowed in URIs/IRIs), and of course the space, but not the '/':
> 
>      #/Bjørn%5C/Carsten%5C/foo%20bar
> 
> Now I really have no clue why surrogate pairs have to be escaped. Why 
> not just use the character? This one didn't show in my mailer or 
> browser, but one would assume that mostly the ones that actually show, 
> at least in some places, will get used. So the actual pointer would be:
> 
>      /Bjørn\/Carsten\/foo 𝄞
> 
> In a fragment (IRI), that would look like:
> 
>      #/Bjørn%5C/Carsten%5C/foo%20𝄞
> 
> If an URI fragment is needed, use UTF-8 and %-encoding:
> 
>      #/Bj%C3%B8rn%5C/Carsten%5C/foo%20%F0%9D%84%9E
> 
> Regards,    Martin.
> _______________________________________________
> apps-discuss mailing list
> apps-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/apps-discuss