[Haskell-cafe] Functional GUIs again
Fraser Wilson
.. not what I expect from a beautiful program, and because what I often
need is a GUI that manipulates a state, and because I don't understand
arrows, and having been intrigued by a recent cafe thread, I threw together
a prototype GUI library with the following features:
- the GUI is defined on three levels: gadgets, widgets and styles
- gadgets are functions on a state
- widgets are data structures which define the layout
- styles are ways to modify the appearance of widgets
The following quick example shows a text box and a button. Clicking on the
button reverses the text in the text box.
> module Main where
> import Barrie
> demoWidget :: Widget
> demoWidget = vbox [ui "demo entry" textBox,
> ui "demo command" (labelButton "click me")]
> type DemoState = String
> type DemoGadget = Gadget DemoState
> demoGUI :: DemoGadget
> demoGUI = localG "demo gui" [editorG "demo entry" id const,
> commandG "demo command" reverse]
> main = gtkMain demoGUI demoWidget "Hello, world"
Two gadgets are used:
editorG :: String -> (a -> b) -> (b -> a -> a) -> Gadget a
commandG :: String -> (a -> a) -> Gadget a
The editor gadget can show a value from a state, and update a state with a
value. The command gadget can transform a state to a new state. gtkMain
connects the gadgets to a widget, which specifies layout using the vbox,
attaching the editor gadget to a text box, and the command gadget to a
button.
Well, that's all pretty trivial. The key thing for me was that I can easily
slap a GUI onto the the front of a class of applications, which happen to be
the sort of applications I've been writing lately. Also, arbitrary parts of
the GUI can respond to things that happen miles away, without really having
to worry about it too much. In barrie-0.1 and 0.2, which used stream-based
approaches, the problem of getting state from one end of the application to
the other was non-trivial.
I'll sketch another quick example:
> data BridgeGame = ...
And a bunch of things you can do with the state:
> makeBid :: Bid -> BridgeGame -> BridgeGame
> playCard :: Card -> BridgeGame -> BridgeGame
For bidding, each bid is represented by a gadget:
> bidG :: Bid -> Gadget BridgeGame
> bidG bid = enabled (bidOK bid) $ CommandG (show bid) (makeBid bid)
'enabled' switches the gadget on if its first argument returns true when
applied to the current state. However, the decision about what to do with a
disabled gadget is made by its corresponding widget.
We get one button for each bid:
> biddingG :: Gadget BridgeGame
> biddingG = localG "bidding" (map bidG allBids)
And they can be displayed in any old order using a widget:
> biddingW :: Widget
> biddingW = vbox (map suitBids [Club, Diamond, Heart, Spade] ++ [ntBids])
> where suitBids suit = hbox $ map (bidButton . flip Bid suit) [1 .. 7]
> ntBids = hbox $ map (bidButton . NT) [1 .. 7]
> bidButton bid = ui (show bid) $ labelButton (show bid)
(You're right, double, redouble and pass are not represented. They make the
lines too long).
Screenshot here: http://thewhitelion.org/images/4D.png
I've just bid four diamonds, so everything lower than that is automatically
disabled.
Currently, Barrie implements buttons, text boxes, labels,
vertical/horizontal layout, single-column lists and drop lists. It current
uses Gtk2hs for rendering, but it's GUI-agnostic (in fact, the first
renderer was putStrLn/getLine).
You can have a look by using darcs:
darcs get http://thewhitelion.org/darcs/barrie
Or get the tarball at
http://thewhitelion.org/haskell/barrie-0.3.0-src.tar.gz
One note: this is not intended to be a theoretically sound approach, merely
a way of getting something done quickly. I would expect it to be most
useful in putting a GUI front-end onto an existing application, in
particular, an application that is driven by user actions which update a
state; e.g. a calculator, a bridge game, a 4th edition D&D character creator
(but that leads to a critical mass of nerdiness, so it's off the table for
now)
cheers,
Fraser.
Lennart Augustsson
2009/2/16 Fraser Wilson <blanc...@gmail.com>:
> Since I'm congenitally lazy, and writing a GUI by hand in the IO monad is
> ... not what I expect from a beautiful program, and because what I often
> _______________________________________________
> Haskell-Cafe mailing list
> Haskel...@haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>
>
_______________________________________________
Haskell-Cafe mailing list
Haskel...@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Fraser Wilson
Username requested ...
cheers,
Fraser.
On Mon, Feb 16, 2009 at 8:10 PM, Lennart Augustsson
<len...@augustsson.net>wrote:
Fraser Wilson
the comment.
I've requested a hackage account, so I expect it to be there shortly :-)
cheers,
Fraser.
On Mon, Feb 16, 2009 at 9:12 PM, Ryan Ingram <ryani...@gmail.com> wrote:
> Tiny code-review comment:
>
> > data Style = Style [(String, StyleValue)]
> > deriving (Read, Show)
>
> > instance Monoid Style where
> > mempty = Style []
> > mappend (Style xs) (Style ys) = Style (xs ++ ys)
>
> =>
>
> > {-# LANGUAGE GeneralizedNewtypeDeriving #-}
> >
> > newtype Style = Style [(String, StyleValue)]
> > deriving (Read, Show, Monoid)
>
> Also, put it on hackage! :) It looks pretty cool.
>
> -- ryan
>
> 2009/2/16 Fraser Wilson <blanc...@gmail.com>:
> > Since I'm congenitally lazy, and writing a GUI by hand in the IO monad is
> > ... not what I expect from a beautiful program, and because what I often
Duncan Coutts
> You must have missed the bit about "congenitally lazy" :-)
>
> Username requested ...
See http://hackage.haskell.org/packages/accounts.html
All you need to do is email Ross and ask.
Duncan
Fraser Wilson
in the sense of "OK, Cafe people, treat this as a user name request and step
to it." What I meant was that I have requested a username (via the email),
and once I have an account I'll put it on hackage.
Sorry for the confusion.
Also, a patch that derives the Monoid instance for Style has been pushed.
Let me explain why I love this mailing list. I hadn't really looked at
Monoids, but then their utility fell out of the recent discussion about
whether they should be called Monoids or ... whatever the other name was.
And now I see them all over the place, and I'm a better person for it.
Surely this is too much to be considered actual programming!
cheers,
Fraser.
On Mon, Feb 16, 2009 at 10:40 PM, Duncan Coutts <duncan...@worc.ox.ac.uk
> wrote:
> On Mon, 2009-02-16 at 20:30 +0100, Fraser Wilson wrote:
> > You must have missed the bit about "congenitally lazy" :-)
> >
> > Username requested ...
>
> See http://hackage.haskell.org/packages/accounts.html
>
> All you need to do is email Ross and ask.
>
> Duncan
>
>
Peter Verswyvelen
That's some great hacking you did :-)
What version of GTK2HS did you use? I get various compiler errors when using
the latest GTK2HS 0.10.0.
Cheers,
Peter
2009/2/16 Fraser Wilson <blanc...@gmail.com>
Fraser Wilson
Thanks!
I haven't tried to compile with 0.10.0 but I can guess that the errors
arise from the use of ListStore. I'm not sure what the best approach
is here. Is 0.9.13 over now? If so, then I'll upgrade and fix it.
For now it would certainly make sense to put the dependency into the
cabal file, which I'm ashamed to say never occorred to me.
cheers,
Fraser
Peter Verswyvelen
different in 0.10.0 (better IMHO)
Yes it would be nice if it worked with GTK2HS 0.10.0, but it might be tricky
to support both version, I don't know.
On Tue, Feb 17, 2009 at 12:51 AM, Fraser Wilson <blanc...@gmail.com>wrote:
> Hi Peter,
>
> Thanks!
>
> I haven't tried to compile with 0.10.0 but I can guess that the errors
> arise from the use of ListStore. I'm not sure what the best approach is
> here. Is 0.9.13 over now? If so, then I'll upgrade and fix it.
>
> For now it would certainly make sense to put the dependency into the cabal
> file, which I'm ashamed to say never occorred to me.
>
> cheers,
> Fraser
>
> On 17 feb 2009, at 00:29, Peter Verswyvelen <bug...@gmail.com> wrote:
>
>
>
> That's some great hacking you did :-)
>
> What version of GTK2HS did you use? I get various compiler errors when
> using the latest GTK2HS 0.10.0.
>
> Cheers,
> Peter
>
>
>
> 2009/2/16 Fraser Wilson < <blanc...@gmail.com>blanc...@gmail.com>
>
>> Now that I re-read my email, it looks like I'm saying "Username requested"
>> in the sense of "OK, Cafe people, treat this as a user name request and step
>> to it." What I meant was that I have requested a username (via the email),
>> and once I have an account I'll put it on hackage.
>>
>> Sorry for the confusion.
>>
>> Also, a patch that derives the Monoid instance for Style has been pushed.
>> Let me explain why I love this mailing list. I hadn't really looked at
>> Monoids, but then their utility fell out of the recent discussion about
>> whether they should be called Monoids or ... whatever the other name was.
>> And now I see them all over the place, and I'm a better person for it.
>>
>> Surely this is too much to be considered actual programming!
>>
>> cheers,
>> Fraser.
>>
>>
>> On Mon, Feb 16, 2009 at 10:40 PM, Duncan Coutts <<duncan...@worc.ox.ac.uk>
>> duncan...@worc.ox.ac.uk> wrote:
>>
>>> On Mon, 2009-02-16 at 20:30 +0100, Fraser Wilson wrote:
>>> > You must have missed the bit about "congenitally lazy" :-)
>>> >
>>> > Username requested ...
>>>
>>> See <http://hackage.haskell.org/packages/accounts.html>
>>> http://hackage.haskell.org/packages/accounts.html
>>>
>>> All you need to do is email Ross and ask.
>>>
>>> Duncan
>>>
>>>
>>
>>
>> --
>> <http://thewhitelion.org/mysister>http://thewhitelion.org/mysister
>>
>> _______________________________________________
>> Haskell-Cafe mailing list
>> <Haskel...@haskell.org>Haskel...@haskell.org
>> <http://www.haskell.org/mailman/listinfo/haskell-cafe>
>> http://www.haskell.org/mailman/listinfo/haskell-cafe
>>
>>
>
Fraser Wilson
nice way of supporting both 0.9.13 and 0.10.0, so I'll take the plunge and
do a destructive upgrade. As you say, the new model/view is much nicer (and
of course I shall be stealing liberally from it).
cheers,
Fraser.