For me, most of the complexity would have been dealing with the governing body :)
Seriously though, being independant allowed me to experiment with things like going directly to the internal Storage API from the client which would have definitely been -1'ed by some committers (rightly so as it would have provided a level of endorsement of this type of client access, making refactoring more difficult). On Sat, Jun 4, 2016 at 11:46 AM, James Carman <ja...@carmanconsulting.com> wrote: > How does it add more complexity by having one governing body (the PMC)? > What I am suggesting is that the driver project be somewhat of a subproject > or a "module". It can still have its own life cycle, just like it does now. > > On Sat, Jun 4, 2016 at 12:44 PM Nate McCall <n...@thelastpickle.com> > wrote: > > > It doesnt. But then we add complexity in communicating and managing > > versions, releases, etc. to the project. Again, from my experience with > > hector, I just didnt want the hassle of owning that within the project > > confines. > > > > On Sat, Jun 4, 2016 at 11:30 AM, James Carman < > ja...@carmanconsulting.com> > > wrote: > > > > > Who said the driver has to be released with the database? > > > > > > On Sat, Jun 4, 2016 at 12:29 PM Nate McCall <n...@thelastpickle.com> > > > wrote: > > > > > > > On Sat, Jun 4, 2016 at 10:05 AM, James Carman < > > > ja...@carmanconsulting.com> > > > > wrote: > > > > > > > > > So why not just donate the Java driver and keep that in house? > > > Cassandra > > > > is > > > > > a Java project. Makes sense to me. > > > > > > > > > > > > > > I won't deny there is an argument to be made here, but as a former > > client > > > > maintainer (Hector), current ASF committer (Usergrid) and active > > > community > > > > member since late 2009, my opinion is that this would be a step > > > backwards. > > > > > > > > Maintaining Hector independently allowed me the freedom to release > > major > > > > features with technology that I wanted to use while maintaining > > backwards > > > > compatibility without having to be bound to the project's release > cycle > > > and > > > > process. (And to use a build system that didnt suck). > > > > > > > > The initial concern of the use of the word "controls" is *super* not > > cool > > > > and I hope that this is being fixed. That said, the reality, from my > > > > (external to DataStax) perspective, is that this is not the case. I > > like > > > > the current project separation the way it is and don't feel like > there > > is > > > > any attempt at "control" of the java driver's direction and > > development. > > > > > > > > -Nate > > > > > > > > > > > > > > > -- > > ----------------- > > Nate McCall > > Austin, TX > > @zznate > > > > CTO > > Apache Cassandra Consulting > > http://www.thelastpickle.com > > > -- ----------------- Nate McCall Austin, TX @zznate CTO Apache Cassandra Consulting http://www.thelastpickle.com
