OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

xri-editors message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: RE: [xri-editors] Groups - xri-spec-syntax-section-v1-r1.doc uploaded

	After an initial readthrough, no major problems jump out at me! Good progress.

	Some comments and questions:

1) We had talked about a "reserved", or "specification-defined" namespace character. Is "$" this character? Or are you intending this *just* for versions? If just for versions, I'd like to suggest that it be the general "specification-defined" namespace and "$v" be used to demarcate versoins. In general, I want to make sure the this proposal lines up with our previous decisions and discussion on versions (it appears to be very close, but I need to take a look). 

2) It would be very instructive to give an example of how XRIs meet the syntax design principles. Its not entirely clear to me in some examples how to interpret the words you have down, and I think the best way to disambiguate is to give examples for each principle. For example, I'm not sure what 2.10 is referring to...

3) Are we sure that XRI syntax can "accomodate" XPath/XQuery syntax? It seems that there would have to be at least some escaping going on there. Not sure.  There is a lot of work going on in the w3c on the path/query languages as well. 

4) [Related to #3] You have # and ? as demarcation characters in the relative-xri production rule. This is fine, but it implies that # and ? are not legal XRI characters. Is this whats intended (I'd +1 that, btw. I'm just raising the issue). 

5) We need to mention escaping rules - ie how to do it legally. This is not something we have to invent, we just have to copy it from somewhere. I18N/unicode issues may also be important here. Also, we can avoid some pain by requiring escaping use uppercase characters -  %HH vs. %hh. 

6) I think we have it covered, but we need to call out that the uri-authority sectoin uses domain names. There are special rules for converting 'idns' (domain names expressed with i18n characters) into domain names (and back) that can be resolved with plain DNS. 
The question, which I believe will be something we just have to decide one way or another, is whether we should always convert dns names, even in cross references, according to the IDN framework, when performing a resolution. The option is to only convert the xri's top-level domain-name for the naming authority. For consistency's sake, my vote is that we must convert all the domain names using the IDN conversion procedure, and not the plain unicode conversion specified in the IRI spec (see the IRI spec for references). This means that it should be very easy to detect when a DNS name is being used somewhere in the XRI, at least where there is any chance that it will get resolved. If the consensus agrees with me, then we should confirm that the BNF allows easy detection (from a parsing point of view) of dns names in an XRI with a bunch of cross references.

7) In the URI production for naming authorities (which we reference), they allow user and password fields to be attached to the dns or ip-identified naming authority. Do we want that? I could see reasons either way, actually. 

Thats it for now. 


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]