Re: [Tmapi-discuss] RESULT VOTE: Feature string/IRI for topic type

[Lars H. <he...@se...>]

Hi Lars,

[...]
>> Well, what I'd like to see is that implementors are not absolutely
>> forced to implement IRIs (for whatever reasons). Lars H. has made a
>> proposal how implementors can get over the demand for IRIs, and I  
>> think
>> we should spent some more time on this. Any thoughts?

> I'm kind of confused by this whole discussion, I must admit. The TMDM
> says all locators are IRIs. Why don't we just leave it at that? If  
> there's any reason to use (or allow the use of) URIs instead, I'd  
> quite like to hear them.

The point is that IRIs are not widely supported, i.e. by Java, PHP,
Python etc.

If we use a test from RFC 3986 like the following:

Base: <http://a/b/c/d;p?q>,
Relative IRI: ?y
Result: <http://a/b/c/d;p?y>


We'd force implementations to develop their own IRI implementation
instead of relying on the standard java.net.URI, for example (not sure
how PHP calls it). In RFC 2396 the result would be: <http://a/b/c/?y>,
not <http://a/b/c/d;p?y>


If we'd divide the tests in RFC 3986-tests and RFC 2396-tests and
reuse the (TMAPI 1.0) feature string notation/URI we'd allow
implementations to pass all TMAPI tests without forcing them to
develop a RFC 3986 compatible IRI implementation.

Best regards,
Lars
-- 
Semagia 
<http://www.semagia.com>

<http://www.topicmaps.de/mailinglist/> German Topic Maps mailing list
<http://tinytim.sourceforge.net/> Open Source Topic Maps engine