add ListT monad transformer and relevant instances

[chessai]

Relevant issue: #54

Don't merge yet, some unresolved questions:

1. Should we include the Enumerator part of the ListT API from pipes? I have not done so here.
2. Should we incur a dependency on exceptions to provide MonadCatch and MonadThrow instances for ListT? There seems to already be the desire to incur this dependency on exceptions in issue [Discussion] Should we bring back exceptions instances? #52.
3. Should the name of this be ListT, or should it be something else?

[treeowl]

If we add this, it definitely should be in its own module!

[chessai]

If we add this, it definitely should be in its own module!

Oh yeah, that was another question I had, but forgot to ask in the first comment. I think I agree.

[treeowl]

What's all this do?

[chessai]

define "all this"

[treeowl]

I just mean it's not obvious to me what listen has to do in a stream contact and why this is how it's done. I'd appreciate a comment.

[chessai]

Initially I thought that MonadWriter might be useful to build up a ListT, and then consume it, but I think an actual use of this might be misguided. Composing Writer with ListT might (would?) be bad; Writer pretty much guarantees space leaks, and ListT is meant for streaming. If someone is ever using the MonadWriter instance for something like ListT, they are likely to be doing something poorly.

Not sure if this instance should exist anymore.

[treeowl]

And this?

[treeowl]

Don't forget to revert all the Streaming.Prelude changes.

[chessai]

I thought I did?

[treeowl]

But you did not. git checkout master src/Streaming/Prelude.hs and then commit.

…

On Wed, Oct 3, 2018, 7:04 PM chessai ***@***.***> wrote: I thought I did? — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#71 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABzi_dwk32pInoCnjG2Nn9a8ZDfRY45oks5uhUKMgaJpZM4XFL5u> .

[PierreR]

What about changing the README ?
https://github.com/haskell-streaming/streaming#-5-how-come-theres-not-one-of-those-fancy-listt-done-right-implementations-in-here

[chessai]

What about changing the README ?
https://github.com/haskell-streaming/streaming#-5-how-come-theres-not-one-of-those-fancy-listt-done-right-implementations-in-here

Yes, if this is approved, the README change will be the last commit to occur. Thanks for pointing this out.

[danidiaz]

Should we incur a dependency on exceptions to provide MonadCatch and MonadThrow instances for ListT?

I'm against depending on "exceptions" in the main "streaming" package. I would prefer the package to remain agnostic about exception and resource handling, even at the cost of some inconvenience.

[chessai]

Should we incur a dependency on exceptions to provide MonadCatch and MonadThrow instances for ListT?

I'm against depending on "exceptions" in the main "streaming" package. I would prefer the package to remain agnostic about exception and resource handling, even at the cost of some inconvenience.

Understandable, it would be nice if more people weighed in on this.

[chessai]

I apologise for the spam - last night's outage caused GitHub to report that i was unable to comment, i hit the comment button multiple times. I guess after things got better, all copies of the comment went through.

[masaeedu]

Are the unresolved questions still unresolved?

[chessai]

My current personal answer to the questions is No, No, and No. Maybe others have thoughts?

[danidiaz]

I agree with the triple-no.

The ListT from "pipes" has a Monoid and compatible Alternative and MonadPlus instances that perform concatenation. Perhaps they could be added here too?

[danidiaz]

This 2014 post by Gabriel Gonzalez gives examples of how concatenation-based MonadPlus instances can be used to write simple library-agnostic streaming sources.