Hi Clay, On Tue, 2015-08-04 at 10:08 -0500, Clay Ferguson wrote: > Robert, > Not to turn this email list into a theoretical discussion, but I'll > say one thing, and then we can carry on the conversation privately or > in different forum... > > I think *the* primary reason OSGi has a place in the world, is > because it can make completely incompatible set of things be able to > run together. For example, if I have component A that requires > version B of of some specific class but perhaps B is using an older > component C than the version of C that A is using internally...then a > single classpath cannot ever work. You must have an environment which > gives each component it's own "world (i.e. separate classpath)" or > environment to run in. > > What SpringBoot is all about, on the other hand, is saying let's > design a single set of dependencies (Technology Stack) that are all > *known* to work together (single classloader) on the same versions of > all dependencies in the chain, and eliminate the version conflicting > before it starts, thus eliminating *one of* the problems OSGi solves. > So OSGi is great, but unless I run directly into one of the problems > it solves, I don't need it. Spring already has "Spring Data MongoDB" > and "Spring Data Solr" projects, and I think there should be a > "Spring Data JCR" project also, that is basically JackrabbitOak > packaged similarly to how I do it in meta64.com. That is, basically > Oak dependencies, with a thin layer of spring beans exposing it, and > a bit of AOP for session management, etc.
Sounds pretty much like something we can 'debate' over a beer, after having settled whether vim or emacs is superior. Just kidding, vim is clearly the winner here ;-) Cheers, Robert
