[adding user@....] Sorry for getting on so late...
My take on this is pretty simple: if we can just get any potential dependencies we need with 'puppet apply...' - let's do just that. The reason we are packing groovy - BTW would appreciate the review for BIGTOP-1423 - is that it wasn't available otherwise. The moment we start maintaining derby, we: a) we'll have to maintain for a long time b) someone will ask to add postgres, mysql, etc. I love postgres, but you see my point... I'd say let's rely as much on the distro vendors: distos are commodity. All of them. Including Bigtop. That's why, I think, we started getting away from being a Hadoop distro project, to Bigdata/In-memory OSS stack. Cos On Mon, Feb 16, 2015 at 05:51PM, jay vyas wrote: > Hi folks. Thanks to @RJ for the seed for this in one of our rambling > conversations, also..... My question: > > *** Is it capable/interesting/relevant to bigtop to packaging a broader > spectrum of the apache ecosystem *** Here some (half baked) rationale > for this idea > > 0 - we recently broadened from the scope of "just hadoop". > > 1 - " big data " per se is starting to blend with other layers of the > stack, like messaging, databases. its kind of tricky to draw a line, IMO, > and will be more tricky going forward. > > 2 - other projects (i.e. derby, groovy, ...) are highly synergistic w/ > bigtop might find a great home here, which is OS neutral and has lots of > great tooling . > > 3 - gives use scope to do bigger and more interesting things inside > bigtop, like deploying an entire full stack data driven environment. > > not even sure how feasible this is (maybe could partially be done with > partnerships with collaboration amongst other packaging communities). but > just thought id put the idea out there . > > -- > jay vyas
