On Fri, 25 Nov 2016 12:06:06 +0530 Rahul Muttineni wrote:

data X = A1 { name :: String, d :: Double} | A2 { name :: String, i :: Int} | A3 { name :: String, d1 :: Double, i1 :: Int}

Now you can use `name` directly to get the string component of the different variants.

Hope that helps!

oops, i forgot to mention. i'd like to be able to write my code; x = [ A1 "a1" 2.0, A2 "a2" 3 ] etc... to save myself a lot of typing.

On Thu, 24 Nov 2016 23:09:26 -0800 Jeffrey Brown wrote:

You still can! Using Rahul's solution, that is.

but wouldn't i have to write A1 {name="a1", d="2.0} A2 {name="a2", i=2} etc... ? That would be ok for these simple examples, but for my actual code the field names would not be just 1 or 2 characters. Brian

On Thu, Nov 24, 2016 at 11:08 PM, wrote:

...

On Fri, 25 Nov 2016 12:06:06 +0530 Rahul Muttineni wrote:

...

data X = A1 { name :: String, d :: Double} | A2 { name :: String, i :: Int} | A3 { name :: String, d1 :: Double, i1 :: Int}

Now you can use `name` directly to get the string component of the different variants.

Hope that helps!

oops, i forgot to mention.

i'd like to be able to write my code;

x = [ A1 "a1" 2.0, A2 "a2" 3 ]

etc... to save myself a lot of typing.

_______________________________________________ Beginners mailing list Beginners@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/beginners

On Fri, 25 Nov 2016 08:41:46 +0000 Rein Henrichs wrote:

No, you would not. Record syntax is an addition to, not a replacement for, the form you want to use.

oh, I see. I had tried A1 { "a1" 2.0 } instead of A1 ("a1" 2.0) Brian

On Fri, 25 Nov 2016 10:19:14 +0100 Daniel Trstenjak wrote:

Hi Rahul,

On Fri, Nov 25, 2016 at 12:06:06PM +0530, Rahul Muttineni wrote:

...

data X =   A1 { name :: String, d :: Double} | A2 { name :: String, i :: Int} | A3 { name :: String, d1 :: Double, i1 :: Int}

Now you can use `name` directly to get the string component of the different variants.

It's not recommended to mix record syntax and ADTs, because you can get runtime errors that the compiler can't catch during compile time, like calling:

i (A1 "foo" 3.2)

agreed. I'm doing this as a form of shorthand instead of creating a text file based input and parsing it. Having any sort of type checking is an advantage.

If you're having the same field in all variants, then an other approach might be better:

data A = Ai Int | Ad Double | Aid Int Double

data X = X { name :: String, a :: A }

yes, that would be better. interestingly in my journeys through the intertubes I have not found a single mention of using the "()" syntax in place of the "{fieldname=value, ...}" syntax as the generator. Brian

On Fri, 25 Nov 2016 18:18:21 +0100 Daniel Trstenjak wrote:

Hi Brian,

...

interestingly in my journeys through the intertubes I have not found a single mention of using the "()" syntax in place of the "{fieldname=value, ...}" syntax as the generator.

The '()' isn't part of the 'data constructor' (the usual term for A1/A2/A3), but sometimes needed to help the compiler to make an unambiguous parsing of the expression.

These are all valid expressions:

i $ A1 "foo" 3.2 -- '$' is often used to get rid of parentheses

let a1 = A1 "foo" 3.2 in i a1

oh, i see. everything's a function. thanks! Brian