[LLVMdev] [RFC] Developer Policy for LLVM C API

[Jim Grosbach]

In the context of this thread, the idea is to provide a better, clearer policy and implementation layering within which such an API could be developed. Right now it’s very muddy, which leads to APIs being developed ad-hoc or not at all. Perhaps I’m being overly idealistic, but I suspect an outcome of this whole thing can be that it will actually be easier to define and integrate APIs because it will be much clearer what the expectations are. JIT clients are very high on the list of consumers of these sorts of APIs for sure.

-Jim

> On Jul 20, 2015, at 3:12 PM, Hayden Livingston <halivingston at gmail.com> wrote:
> 
> Like MCJIT but with on request compliation :-)
> 
> I know, Lang isn't ready and that's fine. But it seems like the
> thread's tune is it will never be ready, i.e. the primary authors of
> OrcJIT will not be providing the implementation of a C Interface.
> 
> We're going to use OrcJIT like many others, as a 3rd or 4th tier jit,
> with inline caching and the works.
> 
> On Mon, Jul 20, 2015 at 1:25 PM, Juergen Ributzka <juergen at apple.com> wrote:
>> I guess Lang isn’t ready to commit to C API yet, since the ORC JIT is still
>> in development. Although I am not sure how to expose all the extensibility
>> that the ORC JIT provides.
>> 
>> What is your particular use case? Do you want just an API for all entry
>> point to the ORC JIT or just an API that allows the use of the ORC JIT in a
>> simple and preconfigured way similar to MCJIT?
>> 
>> —Juergen
>> 
>> On Jul 17, 2015, at 3:04 PM, Hayden Livingston <halivingston at gmail.com>
>> wrote:
>> 
>> For example, ORC APIs in C the bindings.
>> 
>> 
> 
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev