[] == [] - Haskell-Cafe - Haskell.org

newer
How to abstract away set...

[] == []

older
raspberrypi and ghci

Rustom Mody

9 Mar 2015 9 Mar '15

10:33 p.m.

My impression is that earlier [] == [] used to give a type error. Now it gives True. So 1. Which instance of == is being used? 2. Is there some option controlling this behavior?

Attachments:

attachment.html (text/html — 280 bytes)

Reply

Sign in to reply online Use email software

Show replies by date

Lyndon Maydwell

9 Mar 9 Mar

10:37 p.m.

This looks like a great opportunity to try out the new holes functionality: [Prelude] λ [] == _what <interactive>:3:7: Found hole ‘_what’ with type: [t0] Where: ‘t0’ is an ambiguous type variable Relevant bindings include it :: Bool (bound at <interactive>:3:1) In the second argument of ‘(==)’, namely ‘_what’ In the expression: [] == _what In an equation for ‘it’: it = [] == _what Looks like it defaults to [Bool]! Correct me if I'm wrong :) - Lyndon On Tue, Mar 10, 2015 at 1:33 PM, Rustom Mody wrote:

My impression is that earlier [] == [] used to give a type error. Now it gives True.

So 1. Which instance of == is being used? 2. Is there some option controlling this behavior?

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

Rustom Mody

10:42 p.m.

On Tue, Mar 10, 2015 at 8:07 AM, Lyndon Maydwell wrote:

This looks like a great opportunity to try out the new holes functionality:

[Prelude] λ [] == _what

<interactive>:3:7: Found hole ‘_what’ with type: [t0] Where: ‘t0’ is an ambiguous type variable Relevant bindings include it :: Bool (bound at <interactive>:3:1) In the second argument of ‘(==)’, namely ‘_what’ In the expression: [] == _what In an equation for ‘it’: it = [] == _what

Looks like it defaults to [Bool]!

Correct me if I'm wrong :)

Thanks Lyndon. But on my (debian testing) ghc I get: Prelude> [] == _what <interactive>:2:7: Not in scope: `_what' [ghc 7.6.3] And my main question is the second one : For pedagogic purposes I want to get the error -- Whats the (family of) related options?

Reply

Sign in to reply online Use email software

Yuji Yamamoto

10:48 p.m.

And my main question is the second one : For pedagogic purposes I want to get the error -- Whats the (family of) related options?

This feature is called "type holes" introduced by GHC 7.8.1. see https://wiki.haskell.org/GHC/Typed_holes in details. 2015-03-10 11:42 GMT+09:00 Rustom Mody :

On Tue, Mar 10, 2015 at 8:07 AM, Lyndon Maydwell wrote:

...
This looks like a great opportunity to try out the new holes functionality:

[Prelude] λ [] == _what

<interactive>:3:7: Found hole ‘_what’ with type: [t0] Where: ‘t0’ is an ambiguous type variable Relevant bindings include it :: Bool (bound at <interactive>:3:1) In the second argument of ‘(==)’, namely ‘_what’ In the expression: [] == _what In an equation for ‘it’: it = [] == _what

Looks like it defaults to [Bool]!

Correct me if I'm wrong :)

Thanks Lyndon. But on my (debian testing) ghc I get:

Prelude> [] == _what

<interactive>:2:7: Not in scope: `_what'

[ghc 7.6.3]

And my main question is the second one : For pedagogic purposes I want to get the error -- Whats the (family of) related options?

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

-- 山本悠滋 twitter: @igrep Facebook: http://www.facebook.com/igrep Google+: https://plus.google.com/u/0/+YujiYamamoto_igrep

Reply

Sign in to reply online Use email software

Lyndon Maydwell

10:51 p.m.

GHC 7.8.3 here. Hmm, on second inspection I believe that the Bool reference is to the top-level expression, not the list items after all. I don't seem to get any defaulting behaviour either when not using GHCi. This seems to indicate that the extended interactive type defaulting is coming into play: https://downloads.haskell.org/~ghc/7.8.2/docs/html/users_guide/interactive-e... I'd guess that this means that the () default is being used. -Wall seems to indicate that this is in-fact the case: [Prelude] λ [] == [] <interactive>:2:4: Warning: Defaulting the following constraint(s) to type ‘()’ (Eq t0) arising from a use of ‘==’ In the expression: [] == [] In an equation for ‘it’: it = [] == [] True Hope this helps. - Lyndon On Tue, Mar 10, 2015 at 1:42 PM, Rustom Mody wrote:

On Tue, Mar 10, 2015 at 8:07 AM, Lyndon Maydwell wrote:

...
This looks like a great opportunity to try out the new holes functionality:

[Prelude] λ [] == _what

<interactive>:3:7: Found hole ‘_what’ with type: [t0] Where: ‘t0’ is an ambiguous type variable Relevant bindings include it :: Bool (bound at <interactive>:3:1) In the second argument of ‘(==)’, namely ‘_what’ In the expression: [] == _what In an equation for ‘it’: it = [] == _what

Looks like it defaults to [Bool]!

Correct me if I'm wrong :)

Thanks Lyndon. But on my (debian testing) ghc I get:

Prelude> [] == _what

<interactive>:2:7: Not in scope: `_what'

[ghc 7.6.3]

And my main question is the second one : For pedagogic purposes I want to get the error -- Whats the (family of) related options?

Reply

Sign in to reply online Use email software

Rustom Mody

11:20 p.m.

On Tue, Mar 10, 2015 at 8:21 AM, Lyndon Maydwell wrote:

GHC 7.8.3 here.

Hmm, on second inspection I believe that the Bool reference is to the top-level expression, not the list items after all.

I don't seem to get any defaulting behaviour either when not using GHCi. This seems to indicate that the extended interactive type defaulting is coming into play:

https://downloads.haskell.org/~ghc/7.8.2/docs/html/users_guide/interactive-e...

I'd guess that this means that the () default is being used. -Wall seems to indicate that this is in-fact the case:

[Prelude] λ [] == []

<interactive>:2:4: Warning: Defaulting the following constraint(s) to type ‘()’ (Eq t0) arising from a use of ‘==’ In the expression: [] == [] In an equation for ‘it’: it = [] == [] True

Hope this helps.

Thats better -- thanks Lyndon! So after some fishing around I find the more pointed warning I am after is -fwarn-type-defaults The docs seem confusing though: https://downloads.haskell.org/~ghc/7.4.1/docs/html/users_guide/options-sanit... Under -fwarn-type-defaults: says *Have the compiler warn/inform you where in your source the Haskell defaulting mechanism for numeric types kicks in.* Wheres the numeric here?

Reply

Sign in to reply online Use email software

Taylor Hedberg

11:25 p.m.

On Mon, Mar 9, 2015 at 8:20 PM, Rustom Mody wrote:

Under -fwarn-type-defaults: says

*Have the compiler warn/inform you where in your source the Haskell defaulting mechanism for numeric types kicks in.*

Wheres the numeric here?

In GHCi, the type defaulting mechanism is extended beyond numerics. See https://downloads.haskell.org/~ghc/7.8.2/docs/html/users_guide/interactive-e... . In this case, the list is defaulting to type [()].

Reply

Sign in to reply online Use email software

Brandon Allbery

10 Mar 10 Mar

4:33 a.m.

On Mon, Mar 9, 2015 at 10:33 PM, Rustom Mody wrote:

My impression is that earlier [] == [] used to give a type error. Now it gives True.

So 1. Which instance of == is being used? 2. Is there some option controlling this behavior?

The instance being used is (), and it's controlled by -XExtendedDefaultRules. https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/interactive-... -- brandon s allbery kf8nh sine nomine associates allbery.b@gmail.com ballbery@sinenomine.net unix, openafs, kerberos, infrastructure, xmonad http://sinenomine.net

Reply

Sign in to reply online Use email software

Rustom Mody

9:18 a.m.

On Tue, Mar 10, 2015 at 2:03 PM, Brandon Allbery wrote:

On Mon, Mar 9, 2015 at 10:33 PM, Rustom Mody wrote:

...
My impression is that earlier [] == [] used to give a type error. Now it gives True.

So 1. Which instance of == is being used? 2. Is there some option controlling this behavior?

The instance being used is (), and it's controlled by -XExtendedDefaultRules.

https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/interactive-...

Thats what I first thought Tried starting ghci with -XNoExtendedDefaultRules but it does not turn it off

Reply

Sign in to reply online Use email software

Brandon Allbery

9:24 a.m.

On Tue, Mar 10, 2015 at 9:18 AM, Rustom Mody wrote:

On Tue, Mar 10, 2015 at 2:03 PM, Brandon Allbery wrote:

...
On Mon, Mar 9, 2015 at 10:33 PM, Rustom Mody wrote:

...
1. Which instance of == is being used? 2. Is there some option controlling this behavior?

The instance being used is (), and it's controlled by -XExtendedDefaultRules.

https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/interactive-...

Thats what I first thought Tried starting ghci with -XNoExtendedDefaultRules but it does not turn it off

Looks like -X only affects compiler flags: Prelude> :showi language base language is: Haskell2010 with the following modifiers: -XNoMonomorphismRestriction -XNoDatatypeContexts -XNondecreasingIndentation -XExtendedDefaultRules Probably need to use :seti inside of ghci, or in a .ghci / ghci.ini file. -- brandon s allbery kf8nh sine nomine associates allbery.b@gmail.com ballbery@sinenomine.net unix, openafs, kerberos, infrastructure, xmonad http://sinenomine.net

Reply

Sign in to reply online Use email software

Rustom Mody

9:52 a.m.

On Tue, Mar 10, 2015 at 6:54 PM, Brandon Allbery wrote:

On Tue, Mar 10, 2015 at 9:18 AM, Rustom Mody wrote:

...
On Tue, Mar 10, 2015 at 2:03 PM, Brandon Allbery wrote:

...
On Mon, Mar 9, 2015 at 10:33 PM, Rustom Mody wrote:

...
1. Which instance of == is being used? 2. Is there some option controlling this behavior?

The instance being used is (), and it's controlled by -XExtendedDefaultRules.

https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/interactive-...

Thats what I first thought Tried starting ghci with -XNoExtendedDefaultRules but it does not turn it off

Looks like -X only affects compiler flags:

Prelude> :showi language base language is: Haskell2010 with the following modifiers: -XNoMonomorphismRestriction -XNoDatatypeContexts -XNondecreasingIndentation -XExtendedDefaultRules

Probably need to use :seti inside of ghci, or in a .ghci / ghci.ini file.

Ok that did it -- Thanks Brandon. Also thanks Taylor for the exact pointer to the divergence between ghc and ghci semantics

Reply

Sign in to reply online Use email software

3727

Age (days ago)

3727

Last active (days ago)

Download

10 comments

5 participants

tags

participants (5)

Brandon Allbery
Lyndon Maydwell
Rustom Mody
Taylor Hedberg
Yuji Yamamoto