Breaking changes about sparqlee

[jitsedesmet]

Oh, I'm really sorry to hear that. I'm out for the week so I think it would be best to revert the changes until then. I will take a closer look at the issues when I'm back. I must have done something wrong, sorry!

[jitsedesmet]

A does the mentioned commit work? It seems like it's the commit where we introduced the new cache. Maybe I needed to add this module dependency somewhere and didn't?

[rubensworks]

No, that one doesn't work. I'll just revert (as I have to prepare a new release ASAP)

[jitsedesmet]

With even more pain in my hearth I suggest rolling back further than this then, to the commit before we touched the type system since the new type system without a cache is way to slow.

[rubensworks]

Up until what commit would you suggest then?

[jitsedesmet]

Since I'm one my phone it's pretty hard to get the reference but here #112 (comment) I mentioned a few commits, the commit with lowest execution time is the one we want. This is where we introduced the overloadTree but didn't touch the type system.

[jitsedesmet]

I think it was this commit jitsedesmet@c4959d6. Again, I'm really sorry for all this.

[rubensworks]

Ah yeah, that's the one I was referring to in the issue description :-)
So I'll just revert to right before that one.

[rubensworks]

HEAD state at the time of writing is now in https://github.com/comunica/sparqlee/commits/reverted/jitse, and master has been reverted.