,
5 tweets,
2 min read
Read on Twitter
@briman1234 Hmmm. First, of course, yes: any branch I can do without, I will. One way to reduce the actual "switch-time" is to use interfaces and polymorphism. In the various schemes, you "if" once and set an interface delegate to some implementation that will "do the right thing"...
@briman1234 ... This is a good technique, first because it localizes your decision to exactly one point, making it easier/faster to read/write/test/debug.
@briman1234 Second, because a great many conditions for branching that one finds in real apps are really "bulk thens": one condition, many then's applied at various points.
@briman1234 At the same time, a lot of the challenges of branching are unavoidable, and even use of a strategy-like pattern can't quite pull that off. If you make two paths, you want to test the condition *and* you want to test the two paths.
@briman1234 Reducing the number of times you check the condition is great. But it doesn't reduce the number of times the two path-implementors are called, or their need for balance and correctness.
