On 26/02/13 20:16, Joshua TAYLOR wrote:
INSERT { <_:-1ae775c2:13d1823b447:-7ffa> <urn:p>
<_:-1ae775c2:13d1823b447:-7ff9> } WHERE {}
update:
INSERT {
_:b0 <urn:p> _:b1 .
}
WHERE
{ }
(catching up slowly ...)
INSERT with <_:...> bnodes will not work in the way you expect.
The template part of the INSERT is instantiated with new/different
bNodes every time it is used (once per row of the WHERE clause). Here,
there is one row from the WHERE clause. That is not the same as simply
using the INSERT template.
INSERT { ... } WHERE {}
is not the same as
INSERT DATA { ... }
Its a feature of INSERT{} that all bnodes inserted are new (c.f.
CONSTRUCT). That's why SPARQL 1.1 adds the BNODE function that
generates bNodes in the pattern so that, if used carefully, you can
insert new data with bnodes that span result rows.
Andy
