#4191: Replace -fstrict-dicts with programmer-custom strictness annotations for
class contexts
---------------------------------+------------------------------------------
Reporter: LouisWasserman | Owner:
Type: feature request | Status: new
Priority: normal | Milestone:
Component: Compiler | Version: 6.12.3
Keywords: | Difficulty:
Os: Unknown/Multiple | Testcase:
Architecture: Unknown/Multiple | Failure: None/Unknown
---------------------------------+------------------------------------------
Comment(by LouisWasserman):
Here's what I imagine should happen when I have a stack of dictionaries,
each depending on the next. (Let's assume for the moment that there are
no cycles, or at least, that we have chosen to break the cycle at some
particular point.)
(By stack of dictionaries, I mean "in the Core expression that generates
the dictionary, I have a stack of functions of the form f (g (h (...))),
or something of the sort -- you get the idea.)
I'm working on a generalized trie library, and I'm getting non-recursive
dictionary stacks of size 30 when I try to auto-generate trie instances,
and even by hand I get monstrosities like
Maybe `O` Either ((,) Word `App` Bs a)
`O` (K0 Word :*: K0 Word :*: I0 :*: I0)
That type requires 17 dictionaries -- pretty much one for each token in
that string -- and as it stands, I essentially have to rebuild each of
them in order to perform any method at all. Normally, I think this sort
of thing would be done with specialization, but that's not applicable
here, given the range of possible iterations.
A simple solution would be to desugar (instance context => Foo a) as
forcing the dictionaries in the context, modulo dependency loops. That
way, the dictionaries get built exactly once and never again, reducing the
overhead significantly.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4191#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
