catenate / pfsh Goto Github PK

The tests for the paper should test perl's map, reduce (use its module without the module's inefficient functions), and my implementation of filter. The paper should include my implementation of filter, as well as perl's map and reduce.

Since Haskell published its map implementation, add it to the paper as well, for more context around map. Foldr (foldl?), unfold, and zipWith are already there. Add a test to exercise the implementation.

Write a paper or post on why writing for loops should be considered harmful, and why a composition of map, reduce, filter, etc. is better.

The scope of a for loop has access to too much of the surrounding context, so it's difficult to understand and extract, and it's difficult to make sure that variables set in the for loop remain in the for loop (unless the language has variables scoped to the inside of the for loop, but even then you must remember to define them there.

Explicit name and parameters clearly define the interface and summarize a purpose of the for loop body, summarizes the purpose in the body of each argument taken from the context of the call, and checks that the right number and type of parameters are available (this can be done within the for loop, but most programming languages define functions in a way that makes this more explicit and does checks for you).

Separating a for loop body into a function makes it much easier to reuse, and capturing slight differences in the function body is better than copy and tweak since the rest of the body is not duplicated, and the slight differences may be a code smell.

Finally, the body of a for loop may mix and reimplement many patterns of operation over the collection (eg, map, reduce, filter), which would be better expressed by composing them explicitly.