Thanks, I know if using “let”, there’ll be no error. My question is: 1. what does option "—with-ghc” do? 2. Is it better to remove “—with-ghc” when running in intero REPL? If yes, how to remove it?

On May 24, 2017, at 12:22 AM, baa dg wrote:

try with "let": let f x y = x + y

2017-05-23 17:04 GMT+03:00 乃潇 张 mailto:zhangnaixiao@me.com>: I find if I run “stack ghci intero” (remove the “—with-ghc”) from terminal, then the issue will not occur. Does anyone have idea why?

...

On May 23, 2017, at 9:58 PM, 乃潇 张 mailto:zhangnaixiao@me.com> wrote:

Hi, I am trying to define some functions in a REPL window of emacs intero. However, I always got "parse error on input ‘=’” as blow:

Starting: stack ghci --with-ghc intero "--docker-run-args=--interactive=true --tty=false" --no-build --no-load --ghci-options -odir=/Users/spike/Code/haskell/learn-haskell/baby/.stack-work/intero/intero445zea --ghci-options -hidir=/Users/spike/Code/haskell/learn-haskell/baby/.stack-work/intero/intero445zea baby Intero 0.1.20 (GHC 8.0.2) Type :intro and press enter for an introduction of the standard commands.

addTwo x y = x + y

<interactive>:2:12: error: parse error on input ‘=’ Perhaps you need a 'let' in a 'do' block? e.g. 'let x = 5' instead of 'x = 5’

I did some research, seems like “let” is not a must here for ghci version above 8.0. And my ghc version is 8.0.2. So is it a issue of ghci or intero?

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

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

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

On Thu, May 25, 2017 at 06:10:27PM +0300, Baa wrote:

fn :: [a] -> [a] fn lst = do el <- lst guard $ condition el ... return $ change el

How can I do the same but with possibility to call "tell" of "Write" monad in the fn's body? As I understand it should be:

Should be as easy as: import Control.Monad.List import Control.Monad.Writer type Prova = ListT (Writer String) Int fn :: Prova -> Prova fn lst = do el <- lst guard (rem el 2 == 0) lift $ tell "hey bby" return (el + 1)

On Thu, May 25, 2017 at 07:02:58PM +0300, Baa wrote:

В Thu, 25 May 2017 17:43:49 +0200

...

Should be as easy as:

import Control.Monad.List import Control.Monad.Writer

type Prova = ListT (Writer String) Int

Yes! And this is the source of my questions.. 1) How to call it? 2) What does mean to provide argument of type ".. Writer .."? As result, it's good: you can run it with "runWriter", but what is "writer" in input argument? Fake writer (which is ignoring)?

I am not sure for question #1. If we look at the constructor of ListT (`runListT :: m [a]`) and MonadWriter (`(a, w)`) and we compose the two we get: ([a], w) -- or ([Integer], String) in our example -- e.g. ListT (writer ([1,2], "ciao")) <-- to construct No idea what is the correct name of this critter. As we can see by running it, the argument in `fn` (both the list part and the string part) is not ignored: λ> fn $ ListT (writer ([1,2], "ciao")) ListT (WriterT (Identity ([3],"ciaohey baby"))) Does this answer your questions?

ListT is a bit weird in that it affects whatever monad is underneath it, so the order of your types in your Transformer stack matters. Both ways have different meanings and each have legitimate uses. In any case you must use the lift function to get to the monad below the one you are at. import Control.Monad.List import Control.Monad.Writer test :: IO () test = do (runListT $ runWriterT proc1) >>= print (runWriterT $ runListT proc2) >>= print return () proc1 :: Monad m => WriterT String (ListT m) Int proc1 = do tell ("started: " :: String) x <- lift $ ListT (return [1,2]) y <- lift $ ListT (return [3,4,5]) lift $ guard (y /= 5) tell ("x:" ++ show x) tell ("y:" ++ show y) return (x * y) proc2 :: Monad m => ListT (WriterT String m) Int proc2 = do lift $ tell ("started: " :: String) x <- ListT (return [1,2]) y <- ListT (return [3,4,5]) guard (y /= 5) lift $ tell (" x:" ++ show x) lift $ tell (" y:" ++ show y) return (x * y) On Thu, May 25, 2017 at 11:10 AM, Baa wrote:

Hello, everybody!

I can process list in monad style with "do" syntax and to use "guard" function in the body. Something like:

fn :: [a] -> [a] fn lst = do el <- lst guard $ condition el ... return $ change el

How can I do the same but with possibility to call "tell" of "Write" monad in the fn's body? As I understand it should be:

ListT (Writer w) Int

for this example?

- but how to write it? - how to call (run) it? - and how is it safe ("transformers" package has bug in ListT, so "mtl" must be used?)? - is there other canonical way to do it without to use fold*, recursive calls/fix, State/RWS ?

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

To start all these types with T at the end are transformers. They are a type that is wrapped around some inner m. StateT s m, ErrorT e m a, and so on. In order to use do notation, you must be in a type which is an instance of Monad. newtype ListT (m :: * -> *) a = ListT {runListT :: m [a]} instance [safe] Monad m => Monad (ListT m) newtype WriterT w (m :: * -> *) a = WriterT {runWriterT :: m (a, w)} instance [safe] (Monoid w, Monad m) => MonadWriter w (WriterT w m) These types and their instances say the following: ListT m a is a Monad if m is a Monad. WriterT w m a is a Monad if m is a Monad and w is a Monoid. So in order to use do notation in a WriterT String (ListT m) a, I must add the Monad m contstraint to proc, and also ensure that the writer's w is a monoid (it is because it is a string). Now to pass in a ListT as an argument, I must construct one. Remember that in order to use the return function, m must be in a monad, so I must add the Monad constraint. foo :: Monad m => ListT m Int foo = ListT (return [1,2,3]) test = (runListT $ runWriterT (proc3 foo)) >>= print proc3 :: Monad m => ListT m Int -> WriterT String (ListT m) Int proc3 foo = do tell ("started: " :: String) x <- lift foo y <- lift $ ListT (return [3,4,5]) lift $ guard (y /= 5) tell ("x:" ++ show x) tell ("y:" ++ show y) return (x * y) As you saw in the other comment in this thread, most people use a type alias to make it more palatable. type MyApp m a = WriterT String (ListT m) Int -- or type MyApp a = WriterT String (ListT IO) Int proc3 :: Monad m =>ListT m a -> MyApp m Int -- or proc3 :: ListT m a -> MyApp Int On Thu, May 25, 2017 at 12:11 PM, Baa wrote:

В Thu, 25 May 2017 11:52:01 -0400 David McBride пишет:

Hello, David! Am I right that "WriterT ... ListT" is "list of writers"? As I understand, internal representation is "m (a, w)" where m is a-la List? So, this is list of "writers"? I am confused only of this "m" in your "proc1" function, because I suppose this must be Identity and type becomes "WriterT String [Int]" ? Or?

Can this function "proc1" be modified in the way to get input list and to "iterate" over its elements with "do el <- ..." but to can call Writer's tell in the same time? This is the problem for my mind - I can not understand how to pass input list and to have writer inside :) You call ListT's bind but over internal hardcoded list values...

...

ListT is a bit weird in that it affects whatever monad is underneath it, so the order of your types in your Transformer stack matters. Both ways have different meanings and each have legitimate uses. In any case you must use the lift function to get to the monad below the one you are at.

import Control.Monad.List import Control.Monad.Writer

test :: IO () test = do (runListT $ runWriterT proc1) >>= print (runWriterT $ runListT proc2) >>= print return ()

proc1 :: Monad m => WriterT String (ListT m) Int proc1 = do tell ("started: " :: String) x <- lift $ ListT (return [1,2]) y <- lift $ ListT (return [3,4,5]) lift $ guard (y /= 5) tell ("x:" ++ show x) tell ("y:" ++ show y) return (x * y)

proc2 :: Monad m => ListT (WriterT String m) Int proc2 = do lift $ tell ("started: " :: String) x <- ListT (return [1,2]) y <- ListT (return [3,4,5]) guard (y /= 5) lift $ tell (" x:" ++ show x) lift $ tell (" y:" ++ show y)

return (x * y)

On Thu, May 25, 2017 at 11:10 AM, Baa wrote:

...

Hello, everybody!

I can process list in monad style with "do" syntax and to use "guard" function in the body. Something like:

fn :: [a] -> [a] fn lst = do el <- lst guard $ condition el ... return $ change el

How can I do the same but with possibility to call "tell" of "Write" monad in the fn's body? As I understand it should be:

ListT (Writer w) Int

for this example?

- but how to write it? - how to call (run) it? - and how is it safe ("transformers" package has bug in ListT, so "mtl" must be used?)? - is there other canonical way to do it without to use fold*, recursive calls/fix, State/RWS ?

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

---

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

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

If you fill in Identity for m, ListT m Int ends up becoming Identity [Int], which is just a list. Doing effects over a list is a classic FoldM. You can use the foldl library on hackage to do most things, I think. To do it over a list of lists, that actually sounds like something you would use ListT for, but I don't use it much in my code, so I can't say for sure how well it will work for you. For what it is worth I've also heard that the version of ListT in base is not quite correct, although I don't know why. If you are feeling adventurous, check out the pipes tutorial http://hackage.haskell.org/package/pipes-4.3.4/docs/Pipes-Tutorial.html, there is a ListT done right section toward the end where it shows how to construct and use IO code from his version of a ListT. That would require you to understand and buy into the pipes ecosystem. On Thu, May 25, 2017 at 2:37 PM, aquagnu wrote:

David, many thanks for such a detailed answer and explanation!!

So, if I know that I'll iterate over list (vs abstract "m"'s bind), can I replace "m" with Identity, to get list instead of abstract "ListT m"?

And last question: what is more canonical (Haskelish) way to iterate over list with state? Fold*/State/Writer? Actually my ultimate goal was to process list items with some state and possible IO (REST calls, something else)... Is the usage of ListT + Writer (or ListT + State) a good solution or better is to make all logic in one function and to pass it to fold* (so state will be accumulating value)? I will iterate over items of items of this list too: some of them are also lists, so ListT looks more naturally IMHO, but I'm not sure...

David, in all cases - many thanks!!

В Thu, 25 May 2017 12:55:00 -0400 David McBride пишет:

...

To start all these types with T at the end are transformers. They are a type that is wrapped around some inner m. StateT s m, ErrorT e m a, and so on.

In order to use do notation, you must be in a type which is an instance of Monad.

newtype ListT (m :: * -> *) a = ListT {runListT :: m [a]} instance [safe] Monad m => Monad (ListT m)

newtype WriterT w (m :: * -> *) a = WriterT {runWriterT :: m (a, w)} instance [safe] (Monoid w, Monad m) => MonadWriter w (WriterT w m)

These types and their instances say the following: ListT m a is a Monad if m is a Monad. WriterT w m a is a Monad if m is a Monad and w is a Monoid.

So in order to use do notation in a WriterT String (ListT m) a, I must add the Monad m contstraint to proc, and also ensure that the writer's w is a monoid (it is because it is a string).

Now to pass in a ListT as an argument, I must construct one. Remember that in order to use the return function, m must be in a monad, so I must add the Monad constraint.

foo :: Monad m => ListT m Int foo = ListT (return [1,2,3])

test = (runListT $ runWriterT (proc3 foo)) >>= print

proc3 :: Monad m => ListT m Int -> WriterT String (ListT m) Int proc3 foo = do tell ("started: " :: String) x <- lift foo y <- lift $ ListT (return [3,4,5]) lift $ guard (y /= 5) tell ("x:" ++ show x) tell ("y:" ++ show y) return (x * y)

As you saw in the other comment in this thread, most people use a type alias to make it more palatable.

type MyApp m a = WriterT String (ListT m) Int -- or type MyApp a = WriterT String (ListT IO) Int

proc3 :: Monad m =>ListT m a -> MyApp m Int -- or proc3 :: ListT m a -> MyApp Int

On Thu, May 25, 2017 at 12:11 PM, Baa wrote:

...

В Thu, 25 May 2017 11:52:01 -0400 David McBride пишет:

Hello, David! Am I right that "WriterT ... ListT" is "list of writers"? As I understand, internal representation is "m (a, w)" where m is a-la List? So, this is list of "writers"? I am confused only of this "m" in your "proc1" function, because I suppose this must be Identity and type becomes "WriterT String [Int]" ? Or?

Can this function "proc1" be modified in the way to get input list and to "iterate" over its elements with "do el <- ..." but to can call Writer's tell in the same time? This is the problem for my mind - I can not understand how to pass input list and to have writer inside :) You call ListT's bind but over internal hardcoded list values...

...

ListT is a bit weird in that it affects whatever monad is underneath it, so the order of your types in your Transformer stack matters. Both ways have different meanings and each have legitimate uses. In any case you must use the lift function to get to the monad below the one you are at.

import Control.Monad.List import Control.Monad.Writer

test :: IO () test = do (runListT $ runWriterT proc1) >>= print (runWriterT $ runListT proc2) >>= print return ()

proc1 :: Monad m => WriterT String (ListT m) Int proc1 = do tell ("started: " :: String) x <- lift $ ListT (return [1,2]) y <- lift $ ListT (return [3,4,5]) lift $ guard (y /= 5) tell ("x:" ++ show x) tell ("y:" ++ show y) return (x * y)

proc2 :: Monad m => ListT (WriterT String m) Int proc2 = do lift $ tell ("started: " :: String) x <- ListT (return [1,2]) y <- ListT (return [3,4,5]) guard (y /= 5) lift $ tell (" x:" ++ show x) lift $ tell (" y:" ++ show y)

return (x * y)

On Thu, May 25, 2017 at 11:10 AM, Baa wrote:

...

Hello, everybody!

I can process list in monad style with "do" syntax and to use "guard" function in the body. Something like:

fn :: [a] -> [a] fn lst = do el <- lst guard $ condition el ... return $ change el

How can I do the same but with possibility to call "tell" of "Write" monad in the fn's body? As I understand it should be:

ListT (Writer w) Int

for this example?

- but how to write it? - how to call (run) it? - and how is it safe ("transformers" package has bug in ListT, so "mtl" must be used?)? - is there other canonical way to do it without to use fold*, recursive calls/fix, State/RWS ?

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

---

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

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

---

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

-- Best regards, Paul a.k.a. 6apcyk _______________________________________________ Beginners mailing list Beginners@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/beginners

Hello all! If I try to write, for example: instance Alternative MyData where empty = NoMyData a <|> b = if a == b then ... I get error (see on bottom of the mail) sure, bcz I suppose something like: Eq a => (MyData a) All my attempts to add something to instance's `pure` signature have failed. How can I instantiate something with additional restrictions, like in this case? Or are there another solutions for such problem? Interesting is that MyData derives Eq itself! Which, I suppose, must means that "Eq (MyData a)", and if it's true than "Eq a" is true, because how "MyData a" can be Eq without to be "Eq a" (with *automatically deriving* of Eq instance) ?! By the way, for Monoid (which is "* -> Constraint") I can add "Eq a" constraint without problems: instance Eq a => Monoid (Origin a) where mempty = NoMyData mappend NoMyData a = a mappend a NoMyData = a mappend (MyData a) (MyData b)|a == b = MyData a |otherwise = NoMyData but not for Alternative, which is "(* -> *) -> Constraint". *ORIGINAL ERROR DUMP*: ====================== 42 16 error error: • No instance for (Eq a) arising from a use of ‘==’ Possible fix: add (Eq a) to the context of the type signature for: (<|>) :: Origin a -> Origin a -> Origin a • In the expression: a == b In the expression: if a == b then NoOrigin else NoOrigin In an equation for ‘<|>’: a <|> b = if a == b then NoOrigin else NoOrigin (intero) /Best regards Paul

So, if I understood right, I'm trying to restrict instance more than class allows. And only way to accomplish it - is to create own class instead of Alternative which will not be based on Functor :: (* -> *) -> *, like Alternative. No other workarounds? В Wed, 7 Jun 2017 11:25:26 -0400 David McBride пишет:

The Alternative class says nothing about the a in MyData a. It only represents code relevant to MyData.

When you see class Applicative f => Alternative (f :: * -> *), that means all of its functions had to work on any f, where f takes any type and becomes some other type which could be anything.

The reason it works for Monoid is that class Monoid a where implies that a is completely known by the time the instance is fulfilled, therefore the instance can look at what a ended up being and ensure whatever a is, it must have this constraint on it.

You can tell the difference because mempty returns a type that mentions the a mentioned in the class, whereas empty returns an a that is not mentioned in the class, therefore it has to work for any a.

On Wed, Jun 7, 2017 at 8:26 AM, Baa wrote:

...

Hello all!

If I try to write, for example:

instance Alternative MyData where empty = NoMyData a <|> b = if a == b then ...

I get error (see on bottom of the mail) sure, bcz I suppose something like:

Eq a => (MyData a)

All my attempts to add something to instance's `pure` signature have failed. How can I instantiate something with additional restrictions, like in this case? Or are there another solutions for such problem?

Interesting is that MyData derives Eq itself! Which, I suppose, must means that "Eq (MyData a)", and if it's true than "Eq a" is true, because how "MyData a" can be Eq without to be "Eq a" (with *automatically deriving* of Eq instance) ?!

By the way, for Monoid (which is "* -> Constraint") I can add "Eq a" constraint without problems:

instance Eq a => Monoid (Origin a) where mempty = NoMyData mappend NoMyData a = a mappend a NoMyData = a mappend (MyData a) (MyData b)|a == b = MyData a |otherwise = NoMyData

but not for Alternative, which is "(* -> *) -> Constraint".

*ORIGINAL ERROR DUMP*: ====================== 42 16 error error: • No instance for (Eq a) arising from a use of ‘==’ Possible fix: add (Eq a) to the context of the type signature for: (<|>) :: Origin a -> Origin a -> Origin a • In the expression: a == b In the expression: if a == b then NoOrigin else NoOrigin In an equation for ‘<|>’: a <|> b = if a == b then NoOrigin else NoOrigin (intero)

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

---

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

Thank you, David!

I honestly don't think so. But you can ask on haskell-cafe or on stackoverflow if you like. There is a reasonable chance that a class or library exists that already will do what you are looking for, just that I haven't heard of it.

On Wed, Jun 7, 2017 at 12:06 PM, Baa wrote:

...

So, if I understood right, I'm trying to restrict instance more than class allows. And only way to accomplish it - is to create own class instead of Alternative which will not be based on Functor :: (* -> *) -> *, like Alternative. No other workarounds?

В Wed, 7 Jun 2017 11:25:26 -0400 David McBride пишет:

...

The Alternative class says nothing about the a in MyData a. It only represents code relevant to MyData.

When you see class Applicative f => Alternative (f :: * -> *), that means all of its functions had to work on any f, where f takes any type and becomes some other type which could be anything.

The reason it works for Monoid is that class Monoid a where implies that a is completely known by the time the instance is fulfilled, therefore the instance can look at what a ended up being and ensure whatever a is, it must have this constraint on it.

You can tell the difference because mempty returns a type that mentions the a mentioned in the class, whereas empty returns an a that is not mentioned in the class, therefore it has to work for any a.

On Wed, Jun 7, 2017 at 8:26 AM, Baa wrote:

...

Hello all!

If I try to write, for example:

instance Alternative MyData where empty = NoMyData a <|> b = if a == b then ...

I get error (see on bottom of the mail) sure, bcz I suppose something like:

Eq a => (MyData a)

All my attempts to add something to instance's `pure` signature have failed. How can I instantiate something with additional restrictions, like in this case? Or are there another solutions for such problem?

Interesting is that MyData derives Eq itself! Which, I suppose, must means that "Eq (MyData a)", and if it's true than "Eq a" is true, because how "MyData a" can be Eq without to be "Eq a" (with *automatically deriving* of Eq instance) ?!

By the way, for Monoid (which is "* -> Constraint") I can add "Eq a" constraint without problems:

instance Eq a => Monoid (Origin a) where mempty = NoMyData mappend NoMyData a = a mappend a NoMyData = a mappend (MyData a) (MyData b)|a == b = MyData a |otherwise = NoMyData

but not for Alternative, which is "(* -> *) -> Constraint".

*ORIGINAL ERROR DUMP*: ====================== 42 16 error error: • No instance for (Eq a) arising from a use of ‘==’ Possible fix: add (Eq a) to the context of the type signature for: (<|>) :: Origin a -> Origin a -> Origin a • In the expression: a == b In the expression: if a == b then NoOrigin else NoOrigin In an equation for ‘<|>’: a <|> b = if a == b then NoOrigin else NoOrigin (intero)

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

---

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

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

---

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

-- Best regards, Paul a.k.a. 6apcyk

I'll try to clarify the question. Sure I can traverse the stream items with function like `a -> m b` applying it with `mapM`. But in this case I will not have access to return of previous stream node, which I'm planning to use as global state for whole pipe workflow: .--state-------+--state'--------+--state''--> | | | [e0..eN] ==> [e0'...eN'] ==> [e0''..eN''] =====> This "state" will be used for statistics, errors, whatever - through the whold workflow. This "pipe" will iterate over `eN` items (which will be streams too), concatenates results... How this can be achieved with Streaming library? I mean each "node" should have access to stream items but to "global" state (result of prev. node return?) too. Is it possible? В Sat, 27 May 2017 18:30:28 +0300 aquagnu пишет:

I'm trying to start working with Streaming package. First attempt is to simulate source of stream items (I call the function "gen") and some processor ("proc"). Both should be stateful: to be able to sabe some info about processing steps, etc.

... import Streaming import qualified Streaming.Prelude as S ...

gen :: S.Stream (S.Of Int) IO [String] gen = do S.yield 1000 S.yield 2000 x <- lift getLine return ["a", "b", "c", x] -- results

proc :: S.Stream (S.Of Int) IO [String] -> S.Stream (S.Of Int) IO [String] proc str = do e <- str lift $ print "Enter x:" x <- lift getLine return $ e ++ [" -- " ++ x] -- put stream items in result

main :: IO main = do s <- S.mapM_ print $ S.map show gen p <- S.mapM_ print $ proc gen putStr "s: " >> print s putStr "p: " >> print p

And I try to simulate "piping" between "gen" and "proc", seems that function application is enought to compose producers and consumers. But this snippet is not correct: "e <- str" extracts element not from stream, but from results (of "gen"). Even more, I don't know how to "yield" new items, based on stream items. Something like "await" of Conduit, or like in Python "for e in str: ... yield modified(e)...". Is it possible to do it with "do" notation?

--- Best, Paul

-- Best regards, Paul a.k.a. 6apcyk