Back to Questions
1.0k
views8
comments
Feature request (if possible): State-manipulating closures (LISP-style)
f#

Hi! I've just looked through the forum and saw a message with a feature request. I'm new to the forum, but I thought that It could be appropriate to post something similar. Here it is:

In the current version of FSharp, as we know, mutable state cannot be captured in a closure. I do understand that there are some manual methodics for simulating such facility and I also see that this bug/feature is related to OCaml compatibility. I think, however, that most of you will agree that having full-fledged closures in FSharp could be really helpful:

1) much of the FP patterns rely on such closures (AFAIK).

2) those are implemented in many languages like LISP, Ruby, Python, whatever - langugages with non-stack VMs.

3) And the last point - they are implemented even in C#.NET, where the VM is stack-based. That is with the help of preprocessor of course, but FSharp has the preprocessor.

I don't think that such a modification could be radical or painful for most users at the current stage. Either, I do think that this facility could be easily implemeted. And to top all that I'll say that I just beg you for such implementation. Thank you! :)

.

Spammers seems to constantly enlarge their bag of tricks.
I wonder if they use FP for that :-)

By on 5/27/2010 3:27 AM ()

Spammers seems to constantly enlarge their bag of tricks.
I wonder if they use FP for that :-)

Deleted the post you were referring to. Spams do seem to have evolved some new tricks to make there post more reliant, but probably not using FP, since people using FP are inherently good people ;)

By on 5/27/2010 6:23 AM ()

Hi Robert,

If you've got privileges to delete posts, maybe you could delete this one:

[link:cs.hubfs.net]

Looks like link spam to me. I clicked on the "Report Abuse" link on it but I'm not sure if there's a site admin that's watching that stuff or not.

By on 5/27/2010 9:37 AM ()

I think, however, that most of you will agree that having full-fledged closures in FSharp could be really helpful: 1) much of the FP patterns rely on such closures (AFAIK).

What is a full-fledged closure ? Can you give an example that is not doable in F# ?

And most of the FP patterns I learnt are from Haskell which has no mutable state :-)

By on 5/26/2010 5:40 PM ()

To commenters: Yes, I should be more precise when talking about FP patterns. It seems that I've misused the term, and all I wanted to say the state-hiding closures are handy. But that's my fail anyway :)

Again, I'm not implying that F# cannot do FP/Closures in any way. I've just wanted some kind of 'syntactic sugar', not to employ the explicit 'ref' cells, but since in F# creating such references is not verbose and much more consice than in e.g. C#, I think it's better for me to calm down :)

Well, I guess that I'll just be happy with refs. Thank you!

By on 5/27/2010 5:39 AM ()

You may want to read

[link:lorgonblog.spaces.live.com]

to understand the downsides of capturing mutables inside lambdas. It is a very conscious design decision and unlikely to change. I think the way F# is now, it is a little more annoying/ugly, but a lot less error-prone. (Which kinda follows the general principles of an immutable-first language's attitude towards mutability.)

By on 5/26/2010 10:47 AM ()

I've read that, thanks. Still, I'm not sure that functional language must suffer from such limitations. But if that's really just to ensure safety but not to be OCaml-compatible then I might rethink..

By the way, to employ the annoying parallel with Scala - the JVM successor just has it :) I mean mutable state closures.

By on 5/26/2010 10:52 AM ()

If you read Brian's blog post carefully you will learn that you can indeed capture mutable state in lambdas by the means of reference cells. You can argue about the syntactic ugliness of all those ! and :=, but you can hardly qualify this as a language limitation. (Maybe we can take this as a hint from the language designers not to abuse of the feature). You can capture objects too, which may encapsulate mutable state. Finally, are you somehow implying that Haskell is not a functional language ? :)

By on 5/26/2010 11:48 AM ()

    Topic tags

    IntelliFactory Offices Copyright (c) 2011-2012 IntelliFactory. All rights reserved.
    Home | Products | Consulting | Trainings | Blogs | Jobs | Contact Us | Terms of Use | Privacy Policy | Cookie Policy     		Built with WebSharper