Hello Duncan, Monday, August 14, 2006, 2:33:31 AM, you wrote:
Reversing the test doesn't help because either way ghc turns it into:
case thing of _DEFAULT -> 4096 ->
to be exact, ghc passes code for default case separately from code for other cases. look at emitSwitch procedure to know why it's required
and from that generates CMM:
if (thing != 4096) goto much_later; ... much_later: ...
can't you use 'if' expressions?
Even so, I kind of wish there were a stage between STG and CMM that showed the imperative model of STG with linear layout, control flow and notes to indicate thunk/closure allocations. I expect most of my problem is that I do not understand the STG evaluation model sufficiently well to see how it maps to basic blocks, jumps/calls etc.
i once tried to understood STG->CMM code generation. it's all in codeGen directory of ghc sources. and now i think that STG is pretty low-level imperative language, not harder than C for example. i even had the idea to write STG->C translator that generates efficient code. on the other side, jhc already contains pretty the same thing (John claims that jhc's internal language is close to STG) -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com