File: dotnet-lib.html

package info (click to toggle)
hugs98 98.200311-4
  • links: PTS
  • area: main
  • in suites: sarge
  • size: 12,964 kB
  • ctags: 8,084
  • sloc: ansic: 67,521; haskell: 61,497; xml: 4,566; sh: 3,264; cpp: 1,936; yacc: 1,094; makefile: 915; cs: 883; sed: 10
file content (456 lines) | stat: -rw-r--r-- 13,145 bytes parent folder | download | duplicates (7)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN">
<html> <head>
<title>Dotnet library and tools</title>
<link rel=stylesheet href="style.css">
</head>

<body>
<!-- *********************************** -->
<h1>The Dotnet library</h1>

The <tt>Dotnet</tt> library provides you with miscellaneous
auxillary functions to help you interoperate with .NET.

<!-- *********************************** -->
<h2>Representing object references</h2>

<p>
At the base, the library defines and exports the <tt>Object</tt> type
which is used to represent .NET object references:

<pre>
data Object a = ...abstract...

instance Eq (Object a)   where {...}
instance Show (Object a) where {...}
</pre>

The <tt>Object</tt> type is parameterised over a type that encodes the
.NET class reference it is representing. To illustrate how, the
<tt>Dotnet.System.Object</tt> and <tt>Dotnet.System.String</tt>
modules define the following:

<pre>
-- module providing the functionality of System.Object
module Dotnet.System.Object 
         ( Dotnet.Object
	 , module Dotnet.System.Object
	 ) where 

import Dotnet ( Object )

getHashCode :: Object a -> IO Int
getHashCode = ...
...

-- module providing the functionality of System.Xml.XmlNode
module Dotnet.System.Xml.XmlNode where

import Dotnet.System.Object
...
data XmlNode_ a
type XmlNode a = Dotnet.System.Object.Object (XmlNode_ a)
...
foreign import dotnet
  "method System.Xml.XmlNode.get_InnerXml"
  get_InnerXml :: XmlNode obj -> IO (String)

...
-- module providing the functionality of System.Xml.XmlDocument
module Dotnet.System.Xml.XmlDocument where

import Dotnet
import Dotnet.System.Xml.XmlNode

data XmlDocument_ a
type XmlDocument a = XmlNode (XmlDocument_ a)

...
foreign import dotnet
  "method System.Xml.XmlDocument.LoadXml"
  loadXml :: String -> XmlDocument obj -> IO (())
...
</pre>

<p>
[<em>The reason why <tt>Dotnet.</tt> is prefixed to each Haskell
module is to avoid naming conflicts with other common Haskell modules.
See the <a href="#tools">tools</a> for more details.
</em>]

<p>
Notice the type given to
<tt>Dotnet.System.Xml.XmlNode.get_InnerXml</tt>'s argument -- 
<tt>XmlNode obj</tt> -- capturing precisely that the method
<tt>get_InnerXml</tt> is supported on any object that is
an instance of <tt>System.Xml.XmlNode</tt> or any of its 
sub-classes (like XmlDocument.) If we expand a type
like <tt>XmlDocument ()</tt>, we get:

<pre>
XmlDocument () == Dotnet.Object (XmlNode_ (XmlDocument_ ()))
</pre>

<p>
Notice how the type argument to <tt>Dotnet.Object</tt> encodes the inheritance
structure: <tt>System.Xml.XmlDocument</tt> is a sub-class of
<tt>System.Xml.XmlNode</tt> which again is a sub-class of
<tt>System.Object</tt>. The unit type, <tt>()</tt>, all the way to the
right is used to terminate the chain and state that the type represent
just <tt>XmlDocument</tt> (but none of its sub-classes.)

<p>
If instead of <tt>()</tt> a type variable had been used, like what was
done for <tt>get_InnerXml</tt>'s argument type, the type is a subtype.
So, if you've got a <tt>System.Xml.XmlNode</tt> or one of its
sub-classes (like <tt>XmlDocument</tt>), you can safely invoke the
method <tt>get_InnerXml</tt> -- the type variable <tt>obj</tt>
permitting the use of any subtype of <tt>System.Xml.XmlNode</tt>.

<p>
This type trick is a good way to safely represent .NET object references
using Haskell's type system. If you're already familiar with the work on
<a href="http://haskell.org/hdirect/">integrating</a> COM with
Haskell, you'll have already recognised that the type encoding used here
mirrors that used for COM interface hierarchies.

<!-- *********************************** -->
<h2>OO-style application</h2>

<p>
To support the syntax for conventional OO-style method invocation, the 
<tt>Dotnet</tt> module exports the following two combinators:

<pre>
infix 8 #
infix 9 ##

( # )  :: a -> (a -> IO b) -> IO b
obj # method = method obj

( ## ) :: IO a -> (a -> IO b) -> IO b
mObj ## method = mObj >>= method
</pre>

Using these, method invocation can be expressed as follows:

<pre>
  l <- str # Dotnet.System.String.lengthString 
  putStrLn ("Length of string: " ++ show l)
</pre>

<!-- *********************************** -->
<h2>Supporting marshaling</h2>

<p>
The main way to bind to the .NET object model is to use <a
href="dotnet-ffi.html">FFI declarations</a>, but the <tt>Dotnet</tt>
library provides an alternate way (which used to be the only way prior
to the integration of .NET interop into the FFI). This route is mainly
provided for backwards compatibility, so unless you have a specific
reason not to employ the FFI route, the next couple of sections
of this document is of limited interest.

<p>
To support the automatic marshaling of values between the .NET and
Haskell worlds, <tt>Dotnet</tt> provides two Haskell type classes:

<pre>
class NetType a where
   arg    :: a -> InArg
   result :: Object () -> IO a

type InArg = IO (Object ())

class NetArg a where
  marshal :: a -> IO [Object ()]
</pre>

Both <tt>NetType</tt> and <tt>NetArg</tt> work in terms of
<tt>Dotnet.Object ()</tt> -- an untyped representation of object
references.
<p>
The following instances are provided:

<pre>
instance NetType (Object a)
instance NetType ()
instance NetType Int
instance NetType {Int8, Int16, Int32}
instance NetType {Word8, Word16, Word32}
instance NetType Bool
instance NetType Char
instance NetType String
instance NetType Float
instance NetType Double
</pre>

In addition to object references, instances also let you convert
to/from the 'standard' unboxed types that the .NET framework provides.
<p>
The <tt>NetType</tt> class takes care of marshaling single arguments
to/from their .NET representations; the <tt>NetArg</tt> deals with
marshaling a collection of such arguments:

<pre>
instance NetArg ()  -- no args
instance NetType a => NetArg a  -- one arg
instance (NetArg a1, NetArg a2) => NetArg (a1,a2)     -- 2 args
...
instance (NetArg a1, NetArg a2, NetArg a3,
	  NetArg a4, NetArg a5, NetArg a6,
	  NetArg a7) => NetArg (a1,a2,a3,a4,a5,a6,a7) -- 7 args
</pre>

<p>
The idea is here to use tuples to do uncurried method application;
details of which will be introduced in the next section.

<!-- *********************************** -->
<h2>Creating .NET objects</h2>

To create a new object, use one of the following actions:

<pre>
type ClassName = String

new    :: ClassName -> IO (Object a)
newObj :: (NetArg a)
       => ClassName
       -> a
       -> IO (Object res)
createObj :: ClassName -> [InArg] -> IO (Object a)
</pre>

To call the nullary constructor for an object, simply use <tt>new</tt>:

<pre>
main = do
   x <- new "System.Object"
   print x -- under-the-hood this calls ToString() on 'x' 
</pre>

<p>
To use a parameterised constructor instead, you can use
<tt>newObj</tt> or <tt>createObject</tt>:

<pre>
newXPathDoc :: String
            -> System.Xml.XmlSpace
	    -> IO (System.Xml.XPath.XPathDocument ())
newXPathDoc uri spc = newObj "System.Xml.XPath.XPathDocument" (uri,spc)

newBitmap :: Int -> Int -> IO (System.Drawing.Bitmap ())
newBitmap w h = createObj "System.Drawing.Bitmap" [arg w, arg h]
</pre>

<p>
<tt>createObj</tt> lets you pass a list of arguments, but you have to
explicitly apply <tt>arg</tt> to each of them. <tt>newObj</tt> takes
care of this automatically provided you 'tuple up' the arguments. 
<p>
<tt>new</tt> can clearly be expressed in terms of these more general
constructor actions:

<pre>
-- 
new cls = newObj cls ()
-- or
-- new cls = createObj cls []
</pre>

<strong>Note:</strong> the reason why both <tt>createObj</tt> and
<tt>newObj</tt>, which perform identical functions, are provided,
is to gain experience as to what is the preferred invocation
style.

<p>
Unsurprisingly, these two different forms of marshaling arguments are
also used when dealing with method invocation, which we look at next.

<!-- *********************************** -->
<h2>Calling .NET methods</h2>

To invoke a static method, use <tt>invokeStatic</tt> or
<tt>staticMethod</tt>:

<pre>
type MethodName = String

invokeStatic :: (NetArg a, NetType res)
	     => ClassName
	     -> MethodName
             -> a
             -> IO res
staticMethod :: (NetType a)
             => ClassName
	     -> MethodName
             -> [InArg]
             -> IO a
staticMethod_ :: ClassName
	      -> MethodName
              -> [InArg]
              -> IO ()
</pre>

<tt>invokeStatic</tt> uses the <tt>NetArg</tt> type class, so you need
to tuple the arguments you pass to the static method:

<pre>
doFoo :: String -> Int -> IO String
doFoo x y = invokeStatic "System.Bar" "Foo" (x,y)
</pre>

<tt>staticMethod</tt> uses a list to pass arguments to the static
method, requiring you to apply the (overloaded) marshaling function
first:

<pre>
urlEncode :: String -> IO String
urlEncode url = staticMethod "System.Web.HttpUtility"
                             "UrlEncode"
			     [arg url]
</pre>

Instance method invocation is similar, but of course requires
an extra 'this' argument:

<pre>
invoke :: (NetArg a, NetType res)
       => MethodName
       -> a
       -> Object b
       -> IO res

method :: (NetType a)
       => MethodName
       -> [InArg]
       -> Object b
       -> IO a

method_ :: MethodName
        -> [InArg]
        -> Object a
        -> IO ()
</pre>

For example,

<pre>
main = do
  obj <- new "System.Object"
  x   <- obj # invoke "GetHashCode" ()
  print ("The hash code is: " ++ show (x::Int))
</pre>

<!-- *********************************** -->
<h2>Field access</h2>

As with methods, the <tt>Dotnet</tt> library provides access to both
static and instance fields:

<pre>
type FieldName = String

fieldGet :: (NetType a) => FieldName -> Object b -> IO a
fieldSet :: (NetType a) => FieldName -> Object b -> a -> IO ()

staticFieldGet :: (NetType a) => ClassName -> FieldName -> IO a
staticFieldSet :: (NetType a) => ClassName -> FieldName -> a -> IO ()
</pre>

<!-- *********************************** -->
<h2>Using delegators</h2>

To assist in the interoperation with the .NET framework (the UI
bits, in particular), the <tt>Dotnet</tt> library lets you
wrap up Haskell function values as .NET delegators:

<pre>
newDelegator :: (Object a -> Object b -> IO ())
	     -> IO (Object (Dotnet.System.EventHandler ())))
</pre>

When the <tt>System.EventHandler</tt> object reference is passed to
another .NET method, it can invoke it just like any other
<tt>EventHandler</tt> delegate. When that happens, the Haskell
function value you passed to <tt>newDelegator</tt> is invoked. (The
way this is done under the hood is kind of funky, requiring some
dynamic code (and class) generation, but I digress.)

<p>
To see the delegator support in action, have a look at the UI examples
in the distribution.

<!-- *********************************** -->
<h2>Creating a Haskell-based .NET class</h2>

The <tt>Dotnet</tt> library provides experimental support for
creating new classes that wrap up Haskell IO actions:

<pre>
defineClass :: Class -> IO (Object b)

data Class 
 = Class String		-- type/class name
 	 (Maybe String) -- Just x => derive from x
 	 [Method]

data Method
 = Method MethodName       -- .NET name (unqualified).
	  Bool             -- True => override.
 	  String           -- Haskell function to call.
	  [BaseType]       -- Argument types
	  (Maybe BaseType) -- result (Nothing => void).
</pre>

See <tt>examples/class/</tt> in the distribution for more.

<!-- *********************************** -->
<h2 id="tools">Tool support for interfacing with .NET Framework classes</h2>

One thing that immediately strikes you when looking at the .NET
Framework for the first time is its size. Clearly, it wouldn't be
practical to manually type out Haskell wrappers for each and every
class that it provides.

<p>
To assist in the interfacing to .NET classes, a utility
<tt>HsWrapGen</tt> is provided. Given the name of a .NET class,
it generates a Haskell module wrapper for the class, containing
FFI declarations that lets you access the methods and fields of 
a particular class. See the <tt>dotnet/tools/</tt> directory,
if interested.

<p>
<strong>Note:</strong> Hugs98 for .NET makes good use of the
hierarchical module extension to Haskell, so if you do write /
generate your own class wrappers, you may want to consider putting
them inside the library tree that Hugs98 for .NET comes with.

<p>
To demonstrate where and how, supposing you had written a Haskell
wrapper for <tt>System.Xml.Schema.XmlSchema</tt>, you need to name the
Haskell module just that, i.e.,: 

<pre>
module Dotnet.System.Xml.Schema.XmlSchema where { .... }
</pre>

and place it in <tt>dotnet/lib/Dotnet/System/Xml/Schema/</tt> directory
inside the Hugs98 for .NET installation tree. You can then utilise
the wrapper module in your own code by importing it as

<pre>
import Dotnet.System.Xml.Schema.XmlSchema
</pre>

<p>
To avoid naming conflicts with Haskell's hierarchical library tree, we
prefix each of the .NET specific modules with <tt>Dotnet.</tt>. 

<hr>
<address> 
<!-- hhmts start --> Last modified: Wed Mar 12 19:18:34 Pacific Standard Time 2003 <!-- hhmts end -->
</address>
</body> </html>