3 ms·
To be fair, that could be said about most software projects. Yes, absolutely true. The difference here is between "known unknowns" and "unknown unknowns", to p
by pavlov 11y ago
To be fair, that could be said about most software projects.
Yes, absolutely true. The difference here is between "known unknowns" and "unknown unknowns", to paraphrase a certain renowned military strategist...
With Cordova/PhoneGap, it's easy to get the initial impression that you're dealing with "known unknowns", just the typical web development project pitfalls -- "This is HTML, I know this stuff".
When the "unknown unknowns" hit the fan, it's often pretty late in the development cycle, and so things blow past original estimates even though they were competently planned. It's just that the plan was for web development, not the shifting art of websites-in-drag-pretending-to-be-apps-development.
- jkestner 11y agoI agree with that. With an abstraction layer on top of a native layer, inevitably you have to go below the surface unless you fit the intended use case perfectly. I learned this lesson with the Objective C-Python bridge. Spent needless time mapping differences between the two to the point where I had to learn Objective C anyway.
- wpietri 11y agoI agree, but one minor nit: when major problems crop up "pretty late in the development cycle", this typically means a waterfall development cycle. Waterfall approaches are really only appropriate when there's nothing particularly risky. I plead with the HN audience: never let a manager choose the waterfall model by default. They often happen because they make contract negotiation easy. But that's only because they save all the trouble for later.