-
-
Recently Written
-
Follow me on Twitter
What I'm doing
-
Categories
-
Blogroll
-
Links
Life@Work :: Poet Software (’00-’01)
Company Summary
In the midst of the .COM bubble, I joined Poet. Poet at the time had three main products, the legendary object oriented database (Versant ended up buying it), an electronic catalog management system, and an enterprise content management system — both based on the OODBMS.
Technology Keywords
J2EE, Servlets, JSP, EJB 1.0, XML, BEA Weblogic (including the early cluster version), JRun, Poet OODBMS
Projects
ECS
I joined the ECS (electronic catalog system) team, yet another project with a three-letter acronym and implemented numerous extensions into the existing application. This included a flexible importer toolkit, with let the user deploy adapters for BMECAT, CIF, CSV and a few other formats I don’t recollect into the system.
The most fun part about this was definitely the object oriented database.
This nice application was, however, based on an old prototype and not really built according to the latest and greatest patterns. Instead of a nice MVC, it just garbled HTML code in servlets. Surprisingly it wasn’t maintainable anymore and it had to be re-written from scratch.
Who doesn’t want an appserver?
Well, before re-writing this application, I found myself doing a rather fun project. My CTO thought it would be a great idea to have our own J2EE application server (and who didn’t want that at the time). It happened to be that Andersen Consulting in Madrid had what we seemingly needed — a C++ application server. And it must be easy to just write some Java wrappers around that, right?
I spent the following months in Madrid evaluating this application server, and after numerous bottles of Rioja, I came to the conclusion that it was not feasible (actually, my boss still owes me a crate of champagne, simply because I said this to begin with).
ECS Redux
Following this adventure, we started planning the big re-write with product management and started outlining the architecture / prototyping the applications based on:
- an application server agnostic EJB architecture
- enterprise scalability and “clusterability”
- MVC JSP front-end (based on one of the first Java petstores)
- an Oracle back-end (indeed, all customers had Oracle but no-one wanted the OODBMS)
- SOAP-based APIs
The biggest challenge was writing application-server agnostic code. We opted for Together/J which did this out of the box (XDoclet wasn’t around yet).
I was in charge of the database schema and import functionality, but we were really all involved in every piece of the application.
We got to the point where we had a fully working prototype when the times hit us and the company’s stock price went into the toilet (I still have the option papers somewhere ;). To make shareholders happy, the decision was made to abandon the project and patch the old servlet-based application instead.
My entire team decided to leave the company at that point.
