Peter Munch-Ellingsen

Futhark no longer performs the `nimIdentNormalize` step on the output, so the original names from the C sources should be preserved, and you should now be able to create a...

Yes I've seen something similar while wrapping Gtk. It generates the file properly, but for some reason the macro output is invalid. So if you run it twice so it...

The reason why this doesn't work is that Futhark internally uses these identifiers in a hash table. So the representation that is kept is the most cleaned version (capital letters...

Hmm, that is quite strange. It's not impossible that libclang behaves differently between architectures, but this particular instance of a `char *` should be picked up the same I believe....

This should be fixed now, and since this is mostly a duplicate of #87 I'll close this now. Please try it out and see if it works!

> The other one didn't have anything cached yet. That's a bit suspect. Once Futhark runs it should cache both the `opir_X.json` file and the `futhark_X.nim` file in the Nim...

The reason `inlined` weren't generated to begin with is because it makes no sense when dynamically linking. I don't quite remember if this was a matter of actually finding these...

Hmm, header should remove the declaration. Maybe having both cdecl and header confuses it. Try removing cdecl and see if that helps.

Did you ever get a chance to test this? EDIT: Just did some quick and dirty testing with a hand written Nim file and .h file. I can't get it...

> We need to agree on whether it's `var T` or `out T` and I have the impression that the people arguing for `var T` don't understand the consequences. I'm...