[Haskell-cafe] Troubles understanding memoization in SOE

[Peter Verswyvelen]

Thanks Paul & Paul for the answers. I'll certainly read the paper Paul Liu reported.

I just deleted 100 lines of text which explained my problem in more detail, and while I was explaining it, I answered it myself. Typical. I thought the lambda function that memo1 returns would be called over and over again, and instead of reevaluating the stream from the beginning, it would just return the stream since it is in the cache, but actually it just gets called twice in recursive situations: the first time it evaluates y = f x, stores the thunk in the cache, and returns the thunk, the second time it finds the same thunk in the cache, and then computation of the rest of the stream continues without consulting the cache anymore right? From my clumsy explanation you can see that I'm still thinking too imperative, too eager. Haskell is more lazy than I am, which is an incredible achievement :-)

It would really help if I could see the lazy computation; do you think this kind of memo code is traceable using HAT? 

I'll guess I'll have to check out arrows / yampa again. A year ago I did not understand a single thing in those papers, but I should try it again now I read the SOE book :-)

Thanks a lot,
Peter