[einhverfr]

I have found that if I never get the opportunity, chances are it wasn't important anyway. For a lot of things that seem to be moth-balled, they have an amazing way of coming up later.

For example, a month after I gave up on rewriting the financial logic for LedgerSMB, I got a project that required a small subset of rewritten code there. So I went ahead, took the short cuts required for that, wrote an implementation and such in a half a day, from scratch, when my previous attempt took two weeks with nothing to show from it.

That lead to starting the rewrite again which will begin again in earnest after 1.4 branches off. Instead of budgetting months, I am now budgetting only days.

[dclara]

Yes, budgeting for days instead of budgeting for months. When we well manage the task list, usually a feature implementation takes 1-3 days, a project may take 3-4 weeks to reach a milestone. So no feature can take more than a week to finish. Priority is the key.