On 23/05/2016 12:07, Stian Soiland-Reyes wrote:
So after the command line release, perhaps we shall try to assess what is remaining on the server? It should be easier than the workbench, which still has some dependencies and such that needs updating.
The key items (once we've got the command line release) are going to be sorting out data bundling and whether there are any problematic dependencies. We might also want to look at whether there are any parts of the API which don't work and whether we should throw them out or rework them. I don't expect big changes, but a few bits just never grew into what I'd hoped.
On the dependency front, excluding Apache products, server uses Spring core and AOP (which I really hope we won't have to work around, since it uses them very heavily) and DataNucleus for the DAO glue. I don't know what the licensing state of DataNucleus is, but we could work around that as there's really not that much DB interaction. (I originally looked at Hibernate, but found it rather too confusing as it was making all sorts of assumptions that I couldn't match to what I needed to do.)
When understanding the server, realise that it is very declarative. Code generation is the name of the game, whether at compile time or runtime.
Donal.
