Re: [SWObjects-devel] incorrect interpretation of hashes

[Eric Prud'h. <er...@w3...>]

On Aug 13, 2013 2:01 PM, "Ruben Verborgh" <rub...@ug...> wrote:
>
> Hi Eric,
>
> Thanks for the patch!
>
> > I poked around with perl's URI::URL lib and it produces the results
> > expected by Ruben. Attached is a patch which considers …://…#… an
> > absolute based on the "://" and the '#'. I haven't rolled that back
> > into the sparql11 branch yet. Other opinions?
>
> Isn’t it simply: relative if it doesn’t start with a protocol?
>
>
>
> On a different but related note:
> I’m using SWObjects to test graph equality in the EYE reasoner’s Turtle
spec EARL report generator.
> This hash issue was one of the things gave false negatives.
> While the patch solves this, SWObjects only works property on OSX for me.
> If we compile on Linux, we get strange errors like
>
> $ sparql -d IRI_subject.ttl --compare IRI_subject.ttl
> {
>   <http://a.example/s> <http://a.example/p> <http://a.example/o> .
> }
> !=
> +-+
> | |
> +-+

I'll futz with this when I get a chance (ideally this evening when I'll be
near a power plug and downloading the latest boost (can you remind me how
to compile cause I'm sure Ubuntu doesn't have updates for it?)), but what
your seeing is a comparison of two result sets, one of which is a graph and
the other is a table.

> While comparing a file against itself works perfectly on OSX.
> Any clue on what we’re doing wrong with the compilation?
> Or do you perhaps have a compiled Linux version at hand?

I'll have a 64 bit Ubuntu 12.10 when I next connect my laptop. Is that
helpful?

> Best,
>
> Ruben