Developers of right skills want to hand craft every piece of it. This is non-productive. Craftsmanship at the code level shouldn’t be mistaken with efforts you should be putting into product, and code is not product.
It might be if you’re developer but in great sense of any business it usually isn’t. That’s why with TaskBeat we went through the length and actually generate the majority of code. Not all of it yet, but most.
By using the right framework we’ve saved a heck of a development time and can iterate faster than we ever dreamed of. Don’t jump into conclusions just yet. All of it happened through experimentation and the first attempts didn’t lead us far.
Actually the choices we’ve made before TaskBeat was born were limiting us, almost bringing us to a halt. First we’ve made a bad decision of choosing the easy but inflexible framework. Soon we’ve realised that we couldn’t mix the framework of our choice with some other emerging and highly popular frameworks.
It took us great amount of time and even more not to have all the efforts undone before we started again with a better choice of framework. The lessons we’ve learned were used and tested over the course of last two years. The reliance of frameworks to generate code for us brings fantastic results which we can all see now resulting in publishing TaskBeat – one of our greates products online project management among other examples.
When we say code generators we don’t mean we don’t do code. We still do but its not the type of coding you’d do with any other typical web application. We don’t do some low level stuff like routing, combining, wiring things up, etc. Most importantly we don’t do any low level optimisations that can eat up an awful lot of developers’ time. Most of the boilerplate coding, optimisations and memory management is done by a set of frameworks.
We’ve managed to produce a fairly complex application and start focusing on optimising the grand features, not bits and pieces. Some of our developers hate the limited control over every single nut and bolt of code. Most of the more pragmatic developers love the doubling or quadrupling of productivity that comes at the price of control. We’re now ready to push this way of thinking over the last tenth of the project.