Given that the C-binding will be tightly coupled with the C++ layer and written in C++ I don’t think it make sense to have its own repository. In order for the C-binding need to access internal, non-exported methods, for things like serialization it will need to be statically linked to the C++ library. If we use separate repositories and builds then we need to build a static library for the dependency in the C-binding library. This seems overly complicated to me for little to no gain. Can you please elaborate on the the pros of having a separate repository for the C-binding?
For sure agree on the .net core repo. Tough depending on the mixed mode binary support we still might run into the dependency issues on static library. -Jake > On Mar 31, 2020, at 11:23 AM, Matthew Reddington <mredding...@pivotal.io> > wrote: > > I would like to request the addition of two new repositories under Apache in > order to implement this RFC and to take advantage of it in practice. That > would be Apache/geode-c-client and Apache/geode-dot-net-core-client. > >> On Mar 30, 2020, at 3:33 PM, Jacob Barrett <jbarr...@pivotal.io> wrote: >> >> https://github.com/pivotal-jbarrett/geode-native/tree/ee34cfbb5bddb55f5f890bb013c75d7780a787ae/ccache >> >> <https://github.com/pivotal-jbarrett/geode-native/tree/ee34cfbb5bddb55f5f890bb013c75d7780a787ae/ccache> >> >> Quick stab at a POC for thread exception handling. >> >> -Jake >> >
