Re: RDF's relative IRI resolution is ambiguous

> On Sep 3, 2015, at 11:19 AM, Gregory Williams <greg@evilfunhouse.com> wrote:
> 
> On Sep 3, 2015, at 10:36 AM, Ruben Verborgh <ruben.verborgh@ugent.be> wrote:
>> 
>> Hi all,
>> 
>> I've added some more tests at the gist:
>> https://gist.github.com/RubenVerborgh/39f0e8d63e33e435371a
>> 
>> I am now also happy to report that N3.js passes all of them:
>> https://github.com/RubenVerborgh/N3.js/blob/457fa99/test/N3Parser-test.js#L915-L1232
>> 
>> I'd appreciate if you could help verify whether the cases in the gist are indeed correct;
>> if so, then I'll release N3.js 0.4.4 with correct IRI resolution.
> 
> Hi Ruben.
> 
> I’m still concerned that there are issues with some of the tests in the gist. I see you changed some of the tests I mentioned, but didn’t change or comment on others. Some of the new additions are also concerning to me. I haven’t had time to go through these in detail again, but here’s a list of my concerns (with my intuition of what should be produced but isn't):
> 
> Previously mentioned:
> 
> s154 (should not include /ccc/)

This seems to have been corrected.

> New issues:
> 
> s212–s214 (should include /ccc/)

I don’t see that. Base <http://a/bb/ccc/..> has dot segments removed to yield <http://a/bb/>. Joining <g> yields <http://a/bb/c>.

> s218 (should include /ccc/)

> s220–s224 (should include /ccc/)
> s226–s230 (should include /bb/ccc/ or /bb/)
> s238–s251 (should include /bb/ccc/ or /bb/)

Pretty much the same reasoning for all of these.

> Thoughts?

After I made this change I reproduce all of Ruben’s expected output.


> thanks,
> .greg
> 
>

Received on Thursday, 3 September 2015 18:46:34 UTC