Cal addresses that by arranging his book in roughly the order that you should think about things. He recommends that you think about:
1. Version control
2. Issue tracking
3. One-click deploy
4. Internationalization
5. Security
before you start building your application, then you build and release it, then (and only then) start thinking about:
6. Email
7. Web services
8. Scalability
9. Statistics & monitoring
10. APIs
This mostly squares with my experiences (both with my own startup - currently on step 3, with 4, 5 and a launch-ready app already done - and working for others). The only changes I'd make are:
1. Move statistics and monitoring up the priority list, before launch. You want that data available to drive feature implementation.
2. I'm of two minds on internationalization. I think that most apps can wait until they're popular before they need to internationalize. However, i18n is really difficult to do later, after you've already built an app. I watched LiveJournal go through the process, and it wasn't pretty. So even though you don't need it, you may want to do it upfront because it'll be much harder later.
1. Version control
2. Issue tracking
3. One-click deploy
4. Internationalization
5. Security
before you start building your application, then you build and release it, then (and only then) start thinking about:
6. Email
7. Web services
8. Scalability
9. Statistics & monitoring
10. APIs
This mostly squares with my experiences (both with my own startup - currently on step 3, with 4, 5 and a launch-ready app already done - and working for others). The only changes I'd make are:
1. Move statistics and monitoring up the priority list, before launch. You want that data available to drive feature implementation.
2. I'm of two minds on internationalization. I think that most apps can wait until they're popular before they need to internationalize. However, i18n is really difficult to do later, after you've already built an app. I watched LiveJournal go through the process, and it wasn't pretty. So even though you don't need it, you may want to do it upfront because it'll be much harder later.