It is the template, easy templates are able to create fast organizations to discuss crisis fast as you can thought. On resume, you pick a recipe and implement, then at the "certain time" you check what you can change and define this as some sort of innovation. The modules are working, it is just a matter to check how can be refactored under a certain process, without process there is not refactoring. And the debates are the fun part, if you never studied a little bit ORM you don't understand several aspects of the modern computation. And remains discussing strategies to define how to discover if an entity is a new or update, while the world remains coding.
Systems without "a basic help", a very expensive tool to define a model that for year carry a problem under the "schema model visualization" they manage entities but they are not able to understand that "the whole" is different from "a small set of the whole", you cannot move the entity without impact another view perspective. Then you walk with this garbage for like 10years and look manners to "optimize your process". But the most interesting, is that "when you change", you don't exchange six of six, or change the language of your domain expecting "good results on reports".
Today was that very abstract days, you check the machine delegated with tons of useless tools, ports opened everywhere, WIN32 mind, so by the end what do you need is a set of right tools to model, code and deploy, the remaining are mail, infra-structure (another set of tools), security (essentials) and process, like was a discussion to convince why taglibs or scriptlet are not a good choice if you have a computer with 3ghz on client, 2 processors, good monitor and can work with a javascript that will even take 10% of such processing capacity, which means can run with same performance with '30%" of your configuration. "Scale", like you render an HTML page in a server for 10millions of persons in 1hour if very costly, now if your problem even for 1000/h users are already costly, and nobody take such things on place when think on refactor a system, then decide change six for six, like "jstl" and "scriplet" are the same since the "jasper compiler" will act in the jsp at same way.
So if you decide to "put the hand" you offer a "REST alternative", you jump the obsolete already, you don't walk on 10years that already are behind, but the server code and libraries remains the same, same problems same architecture, what changes are the "developer tools", process definition tools, "knowledge base", and for sure, how to write good front ends with rest. Because for example "EXCEL" is a good report tool, just need to be handled by a competent programmer that knows how to page a record on server side and decode a json or xml at client side, no direct connection to lock the DB in production, JQUERY, EXT (nice), GWT, MOOTOOLS, just to begin, the great debate for the lazy programmer is which "component to serialize" a set of pojos on server and how to manage on client, then debate "validation", which means by the end "how to create XML/JSON data structures for a javascript engine. Nobody want to program, "too much complexity", "let for the youth", and the debate, "is it possible to me download the "Mathematica" for complex reports?", or BIRT, JASPER, need to have intelligence "there are others", establish which forms and reports are required to someone spit a string for a browser? By the end, which is obvious, is download a recipe, start to cook and check the results.
There are also, other BI tools, amazing, all of this is due by colony effect, nobody like to think by the end, too much painful.
