Tuesday, August 4, 2009

PHMap and GIPSEPoison

So, I went away for vacation last week and when I got back, a lot of stuff had changed, but I feel it will make for better progress.

First up, I am supporting Chris in making PHMap... which may combine some of my service wrangling code and geo-fetching code with Chris' web code. This is fortuitous because Chris has a much better hang of actual web technologies like CSS, Spring, Javascript, and AJAX... and the concepts that make those all work smoothly together like REST and JSON. Thus, the code should be relatively clean and followable and adjustable, which is much harder when you have things like Java Objects which spitting out Javascript/HTML strings (like in my code).

This also means some of my service-wrangling and geo-fetching logic will probably be broken out or otherwise added-to to better enable the paradigm Chris wants to support . So essentially, refactors are upon us, and it will pretty much be the good old process of splitting function A into functions B and C since Chris will only need functon B.... and updating libraries and setting up property-less jar files for easy downloads instead of complicated builds.

Part of me feels a bit sheepish about how all the code sort of ended up mashed together and a lot of MVC lines were blurred... but then I remember that grid-viewer is essentially an evolution of RODSA-dai, which was a teeny little mashup that would have been overkill to load up with separate controllers and lots of different views. Then, suddenly, we needed polygons, and we needed to interact with the NPDS service, then we needed lots more UI tweaks and the ability to hit different services with different metadata.... and I must have been doing something right because I was able to make those modifications without too many large refactors, it's just now I get to deal with someone else actually needing to use it and not just install it, and their always runs the chance that the two paradigms for how something should work will be a bit obtuse, and larger refactors will be needed to accomodate both of them.

The other thing I will be doing mainly is getting GIPSEPoison working. I implemented a client and several tests and found that it is indeed getting data, but the aggregation needs to be tweaked, and some of the zip3 translations need to be taken into account. I think for that I will just split out some of the code I made for npdsgmaps (so it doesn't have any axis components) and make a new library that can be used by both npdsgmaps and GIPSEPoison.

Otherwise, lots of changes, lots of work, lots of progress, one hopes.

Cheers,
Peter

No comments: