[lldb-dev] A few questions about SWIG and interface compatibility

[Greg Clayton]

We could add the new API in a new "lldb2" namespace to try it out the new API 2.0 which would be available through python as lldb2.*. So the LLDB shared library would vend both the 1.0 API via the "lldb" namespace and the new API 2.0 through the "lldb2" namespace.

So we don't need to necessarily wait to start in on it. We will need to do thorough review as we come up with it...

Greg

> On Feb 26, 2015, at 10:21 AM, Zachary Turner <zturner at google.com> wrote:
> 
> Thanks.  Is there a timeline or roadmap for starting to plan Public API v2.0 (which, if my recollection is correct, will allow us to make breaking changes)?
> 
> On Thu, Feb 26, 2015 at 10:17 AM <jingham at apple.com> wrote:
> 
> > On Feb 26, 2015, at 10:08 AM, Zachary Turner <zturner at google.com> wrote:
> >
> > First, do we require a minimum version of SWIG?  I think the answer to this is currently no.  My next question is whether we can require 3.0?  It was released close to a year ago, so it should be fairly stable.  SWIG 3.0 contains some bugfixes that are useful for generating correct wrappers on Windows, especially with typedefs.
> >
> > My second question is about our interface guarantees.  Are we guaranteeing interface compatibility at the C++ level, or only at the wrapped level?  i.e. is it ok to change the signature of a C++ method as long as SWIG can ultimately generate a wrapper that behaves identically?
> 
> At the C++ level.  We have clients (e.g. Xcode) that use the C++ API's directly.
> 
> Jim
> 
> 
> > _______________________________________________
> > lldb-dev mailing list
> > lldb-dev at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/lldb-dev
> 
> _______________________________________________
> lldb-dev mailing list
> lldb-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-dev