I also would say 2 as we can have a diff, community feedback contributions etc and once we are done we can fork it out to 3.
- Sebastian On 2025/12/05 09:03:52 Christofer Dutz wrote: > Hi all, > > So, I’ve just signed the final paperwork for the NLNet grant and am waiting > for the signed couterpars from them. After that I can formally start > implementing things. > > In general, I will be building a parallel world: > > * > New Read/WriteBuffers > * > New Code Generation > * > New Transports > * > New tools > * > Updated Driver versions. > > Now the question: How should I do that? I see several options: > > > 1. > I create a „java/spi3“ package and add new things parallel on „develop" > 2. > I create a branch and implement things on that > 3. > I create a new repository (As Sebastian probably would prefer and as > discussed in other thread) > 4. > Some option I haven’t thought of ;-) > > 1 has the benefit of allowing a gliding migration, where we should always > have all drivers fully operational during the transition period > Both others result in us having sort of a „split“. > > I personally would prefer option 2. > > Chris > >
