Wed Nov 4 06:28:36 PST 2009 Roman Leshchinskiy
* Consider variables with conlike unfoldings interesting
Ignore-this: ceecbdd10cb85157b1530b98b261e3fe
In this expression:
let x = f (g e1) in e2
the simplifier will inline f if it thinks that (g e1) is an interesting
Tue Nov 3 05:34:31 PST 2009 simo...@microsoft.com
* Document the CONLIKE pragma
Ignore-this: a8e36c2ca7bd43cf327cbb94e1c226d3
Do not merge to 6.12
M ./docs/users_guide/glasgow_exts.xml -10 +54
View patch online:
http://darcs.haskell.org/ghc/_darcs/patches/20091103133431-1287e
b <- [1..y]]
Then you might reasonably argue that it's cheaper to reconstruct xs
for every y, rather than to share it. By the time we've done a bit
of inlining we get
let g = \cn. eftIntFB c n p q
xs = build g
in ...as before...
In short, we'd like to trea
cheaper to reconstruct xs for every
y, rather than to share it. By the time we've done a bit of inlining we get
let g = \cn. eftIntFB c n p q
xs = build g
in ...as before...
In short, we'd like to treat (build g) as CONLIKE, *for this particular g*.
Not f
