I bet most of you know those days where hacking your software goes just fine until you expect a minor feature to cause huge problems. It happened to me on saturday when I wrote the layout manager for my javascript empowered table widget. The architecture will be similar to JTable found in the Java/GNU Classpath API – their idea of the separation between the data model and the view/controller is easy to understand, yet powerful.
Since working with JS means “keep the footprint small”, it also means to rely on browser specific behaviour as much as possible. Guess what! Things did not work as they should! The SUBSEQUENT_COLUMNS-mode messed up the rendered data as it was hopping and popping all over the screen while resizing a column. This morning, 7am, after trying out hundreds of thousands different algorithms to compute the overflow (”spill” called by GNU Classpath, “delta” by the Sun Java API) and share it between the columns right-hand sided to the resizing column, I figured out that an immediate call to [CuTableColumn].setPreferredWidth() right after calling [CuTableColumn].setWidth() helped! It works like a charm now.
I got a job offer by a chinese software company. They are looking for a team leader for their PHP department in Hong Kong. So if you happen to be flexible enough to move to Hong Kong (or better yet, if you live there), know PHP very good and have an authoritarian personality, just contact me, I’ll send your contact data right away. Love my office too much here in ol’ Aachen!