Hello,
I was wondering if there is someone already working on putting the new OrcJIT APIs in the LLVM-C bindings?
Also, is there a general consensus to also add C bindings when new major features are added?
Hayden
Hello,
I was wondering if there is someone already working on putting the new OrcJIT APIs in the LLVM-C bindings?
Also, is there a general consensus to also add C bindings when new major features are added?
Hayden
Hello,
I was wondering if there is someone already working on putting the new
OrcJIT APIs in the LLVM-C bindings?Also, is there a general consensus to also add C bindings when new major
features are added?
Generally not, so far as I know. Things are added to the C bindings on an
as-needed basis, generally.
Thanks, David.
I’d be happy to add the bindings … is there a general way we add them? Or do you just scrub the API and make sensible judgements to the API?
Thanks, David.
I'd be happy to add the bindings .. is there a general way we add them? Or
do you just scrub the API and make sensible judgements to the API?
Pretty much that, I suppose - that's my rough understanding anyway.
Thanks, David.
I'd be happy to add the bindings .. is there a general way we add them?
Or do you just scrub the API and make sensible judgements to the API?Pretty much that, I suppose - that's my rough understanding anyway.
That said, ORC is a relatively new API that may be in flux and the C API
does come with a rather strong (I forget if we've clarified this & made
different guarantees) guarantee of backwards compatibility, so it might not
be appropriate to expose such a detailed API especially for a
new/experimental piece of code - not sure.
HI Hayden,
There's no plan to expose any of the Orc APIs directly just yet - they're still very experimental.
We could add a call to opt-in to using OrcMCJITReplacement rather than MCJIT though.
Cheers,
Lang.