InCAST Update - October 13
Stuart Moffatt
The entire project was folded into one eclipse project for better
builds and less configuration in the development environment. The
project is now under gov-fema-hazus-incast. The server and client
modules were migrated to this namespace.
The build environment was moved from a strict Eclipse / Ant build to
Maven, and then back again once it was discovered that GWT + Maven +
Eclipse don't really play nicely together, especially across a shared
project space as we had before. So while I am targeting eclipse as the
only dev environment for InCAST, I also believe that a simple dev
environment wins. I still like the idea of Maven, but integration into
my IDE was too buggy.
On the Hibernate front I have finished the O/R mappings. There are now
~17 models in all which are reflected as tables in the database. I am
sure that a hibernate guru could poke holes in my config (please do!)
but it is working well right now: I have numerous test cases saving
data properly, and the GWT client side code is saving properly too
(which was problematic before because of recreating a hibernate
session with each serialized hit to the API). The fix was to implement
versions are hibernate suggests. Still to do: test versions against
multi-user locking scenarios and see how it stacks up.
For the Google Web Toolkit (GWT) work, I have been pleasantly
surprised at the flexibility of the framework. I now have a working UI
that allows a user to create a new building, populating its
characteristics across many UI screens tucked into a number of double-
decker tabs (still need to work out the chrome a bit though). What is
fantastic are the click and focus listeners that have made updating
the building model instantaneous. Likewise, the data provider /
updater model is working well. I have had to create some longish code
in trying to make the building model trickle down into the UI object
hierarchy, especially after an async update is complete. But the
pattern is there now, so... unless someone joins and has a better
idea.
Still to come: implementing the building list and search functionality
(or re-implementing, as I did most of the building list already, but
it was native JDBC (very time consuming) and the object hierarchy was
mangled. I am hoping the new ApplicationManager will help me control
things better this time.
It is nice now that I can see the light at the end of the tunnel. It
is hard to imagine I started this project in August and have only been
at it part-time since then. There are tons of features that I want to
role into this project (other FEMA helper tools like BIT and FIT, and
new additions to InCAST like 'building notes' and BaseCamp project
management integration) but I will have to wait until Release 1 is out
the door before I dream about getting other devs on board to start
adding features.
As far as a release goes, I am working full-time until I resume my
studies in January 2008 so I am planning on releasing before then. I
would like to say mid-November, but that is really dependent of the
key part of my job which is helping the creation of mitigation goals
and objectives, and interpreting HAZUS' loss analysis data to inform
that process.