| > Suddenly someone is like “I would have done it from this other approach” and you have no idea whether it's tractable I find myself on the other side of this, all the time. Suddenly someone comes up with some work that they have done without involving anyone else. I know this codebase, and I know moving that cache invalidation will make the codebase harder to understand, and also runs counter to the multi month effort you had to move all cache handling to the shared library which every other codebase in the product uses. I try to be very careful about it, "have you considered using this other approach instead?" usually means unless you do this there will be the need for an even bigger rewrite in the future, but nobody can really take comfort in that. The answer is that post facto code review is really unsuited for collaborative development. You start some work, you better involve other people straight away. Bounce ideas with a partner or two that really knows the codebase so you both understand what should be done and why. Unless you agree what it is your suggested change should achieve, there can be no useful code review! When you are very a tiny team this mostly follows naturally, but always gets lost when the team grows and are split in several, and when individual developers starts to lose track of the codebase as a whole. |