On 2/28/07, Werner Punz <[EMAIL PROTECTED]> wrote:
Arash Rajaeeyan schrieb: > oh yes, also conversation scope of Trinidad > does you (or any one one else) have access to JSR 299 info? > do you which approach are they going to standardize for > conversation/dialog/(or what ever they name it)? > Btw. speaking of JSR 299, and conversations, isnt jsr299 just a glue specification of marrying ejb3 and jsf so that you can use ejb3 beans as managed beans?
JSR299 means whatever the expert group decides to propose to the JCP executive committee for approval. The kinds of things you describe above are definitely within the initial proposal document[1], but I would not place any bets that JSR-299 will limit itself to just what you mentioned.
Regarding conversations and dialogs: This stuff really belongs into the servlet space just like session and request, which technologies then are built on top of such scoping is an entire issue.
Agreed in general ... the devil is in the details. Consider the RESTafarian attitude that scopes of any kind (other than request scope) are evil. And, consider the fact that, although javax.servlet was one of the earliest "extension" proposals for the Java language, there have not been any mainstream-adopted solutions on different APIs to adapt HTTP requests to Java business logic (unless, I suppose, you count SOAP mappings via things like JAX-RPC and JAX-WS ... but those still count IMHO as built on *top* of the servlet APi instead of replacing it). Maybe there is some virtue in a simple baseline standard that everyone can adopt?
Lets face it 90% of all problems most people have in webapps stem from the fact that you cannot keep objects for a longer time without going through the problems a session scope causes for garbage collection and due to the fact that if you do not work on a pure jdbc base but on an orm base you either have to keep an erm open for your entire session with all related problems or you have to open it on request or even works on business case and then run into the usual merge problems most orm layers have (which is not solvable in a satisfying way anyway) The current already big number of various dialog systems which also keep something conversational open for object storage stem from the fact that this is a huge problem or has become a bigger one now that webapps seem to have moved towards orm layers where this problem becomes more problematic. Tackeling it on JEE level has been long overdue in my opinion especially because most of the usage and core patterns basically are tested by now. Craig since you are reading this, any idea if the servlet specs will be opened to scopes/conversations in the next specifications?
It turns out that I've been an internal proponent of dealing with these kinds of issues at the servlet spec level, as my colleagues in the platform group are aware of :-). One of the critical challenges, unfortunately, is econimics -- funding any spec all the way through the JCP process is likely to be a six-figure ($) investment, and the challenge is to optimize our (Sun's) investments. My personal interest in this problem space is actually at a level *below* the servlet API ... hopefully, the recently filed RESTful API JSR can deal with those sorts of things. Things like conversation scope (and even session scope) should be extensions on top of such a basic API, not fundamental features of it. Serv;ets have served us honorably for almost 10 years (only a little less than the lifetime of Java itself) ... but it's time to move forwards.
Werner
Craig
