Joachim Van Herwegen

> Not sure if bug or feature.. The WebID profile spec is not implemented yet in CSS. The spec does also say this > When an app operating on behalf...

At the bottom is a list of changes that are going to be necessary to implement this. The thing is though, I'm not sure any of this is really required...

Ignore my previous message if you got it by mail. The issue is that the code first executes the first part (the minus) and then tries to join the results...

I think they would, probably should run some tests to be sure. Would still be an issue for more complex queries where the `+` part is more internal to the...

As far as I can see I think @thomas-delva is actually right, since the 2 UNION parts are completely separate, the first part corresponds to ```sparql :a :b ?s. BIND(?s...

Could also be based on the internal algorithm? E.g. if there are 100 matches for the first bgp, solving every match completely in the subqueries is 1%.

I only did some minor tests before getting distracted by something else, but first results seemed to indicate: Fastest is the standard Node.js stream if you have a datatype that...

Well they are still sync generators, so you lose the whole async advantage of asynciterator there. The moment you start having streams where you need to wait on data this...

> If the root container serves as a storage, that is the only one; if not, storages are typically created on a single hierarchical level (either as subdomains or as...

That sounds correct. I would modify the function I linked above to take a boolean as input to indicate if this is a read or write request, and throw the...