Mostly I'm looking for a rough estimate. Some false positives and
false negatives are tolerable. If I have something like
f :: Int -> Maybe String -> String
f _ Nothing = "Hi there!"
f n (Just b) = if n > 0 then show b else "whatever"
then I'd likely be interested in a warning about the fact
Ömer Sinan Ağacan writes:
> Wait, isn't new codegen merged years ago? I don't see these comment
> lines in 8.0.1-generated Cmm files. Comments in the AST may be there
> but no comments were printed until very recently.
>
Are you sure you are looking at code generated with
2016-05-29 11:14 GMT-04:00 Ben Gamari :
> CmmUnsafeForeignCall as it looks quite similar to
> a CmmCall
Well then maybe we should print those differently instead of adding
noise to every single line just to distinguish CmmUnsafeForeignCall
from CmmCall.
Wait, isn't new codegen merged years ago? I don't see these comment
lines in 8.0.1-generated Cmm files. Comments in the AST may be there
but no comments were printed until very recently.
If no one feels strongly about this I'd love to remove those.
2016-05-29 11:14 GMT-04:00 Ben Gamari
Hi all,
thank you for the quick response.
Since ImpredicativeTypes is not a road I want to go down, a newtype instead
of a type synonym seems like the best bet for that particular case.
Avoiding impredicativity "by accident" makes complete sense to me. I just
thought to bring up the example on
Richard Eisenberg writes:
> Hi devs,
>
> I'm forwarding the following from a friend (cc'd) who was unable to register
> for Trac. Can anyone help?
>
>
>
> Hi. I tried to register with username "j6carey" and email
> "j6ca...@gmail.com" to report a bug in GHC 8.0.1, and
Hi devs,
I'm forwarding the following from a friend (cc'd) who was unable to register
for Trac. Can anyone help?
Hi. I tried to register with username "j6carey" and email "j6ca...@gmail.com"
to report a bug in GHC 8.0.1, and the system says there's a 97% chance I'm a
spammer. It asked
The non-outer variant works, because then there aren’t higher rank types at
all, i.e. `state` of `Handler` is free to flow outwards.
There is two ways to fix issue: Either use `newtype` or use `ImpredicativeTypes`
—
{-# LANGUAGE RankNTypes #-}
module TestTypes where
data State a=
Hi Michael,
I think that GHC 8 is correct to reject this program without
ImpredicativeTypes (and if you enable ImpredicativeTypes, all bets are
off). Previous versions incorrectly accepted such programs, because they
didn't look through type synonyms. If you expand the type of the second
The same bug has bitten git-annex too. IIRC.
Cheers,
Gabor
Em domingo, 29 de maio de 2016, Michael Karg
escreveu:
> Hi devs,
>
> could you please have a look at the following code snippet (modeled after
> a real-world app of mine)? There's a rank2type involved, and
Hi devs,
could you please have a look at the following code snippet (modeled after a
real-world app of mine)? There's a rank2type involved, and it doesn't
type-check anymore when the type is e.g. part of a tuple, whereas
everything's fine when it's the "outermost" type.
With GHC7.10 both
Ömer Sinan Ağacan writes:
> I'm reading a lot of Cmm these days and comments added by Cmm dump (which are
> apparently added after 8.0.1) are so annoying becuase they're not saying
> anything useful (what's the point of adding "// CmmCall" to a "call" line or
> "//
12 matches
Mail list logo