--On Wednesday, March 02, 2005 12:24 PM +0100 "Schlesinger, Frank-Michael" <[EMAIL PROTECTED]> wrote:
What are the open questions? ----------------------------
- Does my modeling idea make sense in respect to our requirements and the "HackyStat" architecture ? - I mean, is it even possible or did I miss something during my analysis ? - Or is there a better (less invasive, more elegant, ... ) way than the subclassing idea ?
Hi Frank,
Your approach seems reasonable to me. The only additional suggestion I have beyond what I've already stated is that you might want to start with a "spike solution" in which you simply dump out your micro process events directly to the local file system in XML without using Hackystat at all. Using something like JDOM (http://www.jdom.org), you can read/write your events as XML in very few lines of code.
This enables you to quickly jump to the important part of the problem, which is (a) what kinds of micro process events are interesting, and (b) what are the appropriate abstractions of micro process events, and how does one generate them?
Once you've got a handle on those issues, then you can start figuring out where Hackystat fits in.
P.S.: I'm having trouble in subscribing to the hacky mailing lists. The [EMAIL PROTECTED] tells me my mail is forwarded to [EMAIL PROTECTED] to which unfortunately no route exists.
My bad. It should have been '[EMAIL PROTECTED]'. I've fixed the instructions at <http://www.hackystat.org/>. For debugging purposes, if you could try again and let me know if you run into problems, that would be great!
Cheers, Philip
