#5256: code in an "else" branch that is never executed still does influence
runtime (Vectorizer?)
---------------------------+------------------------------------------------
Reporter: j.waldmann | Owner:
Type: bug | Status: closed
Priority: normal | Milestone:
Component: Compiler | Version: 7.0.3
Resolution: wontfix | Keywords:
Testcase: | Blockedby:
Difficulty: | Os: Linux
Blocking: | Architecture: x86_64 (amd64)
Failure: None/Unknown |
---------------------------+------------------------------------------------
Changes (by simonpj):
* status: new => closed
* resolution: => wontfix
Comment:
Thanks for the diagnosis Daniel; you are exactly on target I think.
I don't think there is much we can do automatically here. Yes, foldb_cap
has some static arguments, but SATting it would make no difference because
the resulting function is big. What you want is ''both'' to SAT it ''and''
to INLINE it, so that the entire fuction is specialised to its functional
arguments at each individual call site. If that's what you want it's
probably easier to do it by hand.
The one feature I can imagine being useful would be this. Suppose you add
INLINE to `foldb_cap` in `bug.hs`. It won't have any effect because
`foldl_cap` is recursive. But GHC could treat that pragma as a clue to
SAT the function, so that it ''can'' be inlined.
I'm not sure though. It seems a bit indirect and elaboate. Maybe better
to to be explicit.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5256#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
