www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Reflection would be really nice

reply "Frank Benoit (keinfarbton)" <benoit tionex.removethispart.de> writes:
A little hello world in SWT is actually bigger then 7 MB.
This is due to the generated reflection code, and the need to pull in
the whole SWT in the initialization phase.

With rudimentary reflection, this would be much better.

I also think, there would be many application possible with this kind of
stuff.

class ClassInfo{
	Field[] fields;
	Method[] methods;
	Constructors[] constructors;
	static ClassInfo[] allClasses; // probably the most important
}

class Field{
	uint offset;
	TypeInfo type;
	Attribute[] attribs;
	char[] name;
	void* getPtr( Object* obj );
}
class Parameter{
	char[] name;
	TypeInfo type;
	Attribute[] attribs;
}
class Method{
	uint offset;
	TypeInfo ret_type;
	Attribute[] attribs;
	char[] name;
	Parameter[] params;
	void* getPtr( Object* obj = null );
}
class Constructor{
	Attribute[] attribs;
	char[] name;
	void* getPtr();
	Parameter[] params;
}
Apr 19 2007
next sibling parent reply Walter Bright <newshound1 digitalmars.com> writes:
Frank Benoit (keinfarbton) wrote:

 class ClassInfo{
 	Field[] fields;
 	Method[] methods;
 	Constructors[] constructors;
 	static ClassInfo[] allClasses; // probably the most important
 }


allClasses is already there, see Classinfo.find().
Apr 19 2007
next sibling parent reply "Frank Benoit (keinfarbton)" <benoit tionex.removethispart.de> writes:
 
 allClasses is already there, see Classinfo.find().


Didn't know about it :)

Now i can test if "org.eclipse.swt.SWT.SWT" exists.
If so, is there a way to call the static method "static_this()"?
Apr 19 2007
next sibling parent Sean Kelly <sean f4.ca> writes:
Frank Benoit (keinfarbton) wrote:

 allClasses is already there, see Classinfo.find().


 
 Didn't know about it :)


It was added in 1.011 (IIRC), so that's not surprising :-)


Sean
Apr 19 2007
prev sibling parent reply Walter Bright <newshound1 digitalmars.com> writes:
Frank Benoit (keinfarbton) wrote:

 allClasses is already there, see Classinfo.find().


 
 Didn't know about it :)
 
 Now i can test if "org.eclipse.swt.SWT.SWT" exists.
 If so, is there a way to call the static method "static_this()"?


In the module info, there should be an array of the static constructors.
Apr 19 2007
parent reply "Frank Benoit (keinfarbton)" <benoit tionex.removethispart.de> writes:
Walter Bright schrieb:

 Frank Benoit (keinfarbton) wrote:

 allClasses is already there, see Classinfo.find().


 Didn't know about it :)

 Now i can test if "org.eclipse.swt.SWT.SWT" exists.
 If so, is there a way to call the static method "static_this()"?


 
 In the module info, there should be an array of the static constructors.


I cannot use them, because of circular dependencies. This is, why i
"manually" call "static_this()" methods (note the underscore) in a
defined order.
Apr 19 2007
next sibling parent Walter Bright <newshound1 digitalmars.com> writes:
Frank Benoit (keinfarbton) wrote:

 Walter Bright schrieb:

 Frank Benoit (keinfarbton) wrote:

 allClasses is already there, see Classinfo.find().


 Didn't know about it :)

 Now i can test if "org.eclipse.swt.SWT.SWT" exists.
 If so, is there a way to call the static method "static_this()"?


 In the module info, there should be an array of the static constructors.


 
 I cannot use them, because of circular dependencies. This is, why i
 "manually" call "static_this()" methods (note the underscore) in a
 defined order.


There isn't a way to do that with reflection at the moment.
Apr 19 2007
prev sibling parent "Vladimir Panteleev" <thecybershadow gmail.com> writes:
On Fri, 20 Apr 2007 01:39:12 +0300, Frank Benoit (keinfarbton)
<benoit tionex.removethispart.de> wrote:


 Walter Bright schrieb:

 Frank Benoit (keinfarbton) wrote:

 allClasses is already there, see Classinfo.find().


 Didn't know about it :)

 Now i can test if "org.eclipse.swt.SWT.SWT" exists.
 If so, is there a way to call the static method "static_this()"?


 In the module info, there should be an array of the static constructors.


 I cannot use them, because of circular dependencies. This is, why i
 "manually" call "static_this()" methods (note the underscore) in a
 defined order.


Speaking of circular dependencies, those are quite annoying. Would it be
possible to add a module/static constructor modifier keyword like "standalone"
(doesn't depend on the static constructors of imported modules) or "depends"
(which allows to enumerate the modules this [module's] static constructor[s]
depend on)?

-- 
Best regards,
  Vladimir                          mailto:thecybershadow gmail.com
Apr 20 2007
prev sibling parent reply "Frank Benoit (keinfarbton)" <benoit tionex.removethispart.de> writes:
Walter Bright schrieb:

 Frank Benoit (keinfarbton) wrote:

 class ClassInfo{
     Field[] fields;
     Method[] methods;
     Constructors[] constructors;
     static ClassInfo[] allClasses; // probably the most important
 }


 
 allClasses is already there, see Classinfo.find().


Where can i find the ClassInfo DDoc?
It seems, it disappeared from
http://www.digitalmars.com/d/phobos/object.html
Apr 19 2007
parent torhu <fake address.dude> writes:
Frank Benoit (keinfarbton) wrote:

 Walter Bright schrieb:

 Frank Benoit (keinfarbton) wrote:

 class ClassInfo{
     Field[] fields;
     Method[] methods;
     Constructors[] constructors;
     static ClassInfo[] allClasses; // probably the most important
 }


 
 allClasses is already there, see Classinfo.find().


 
 Where can i find the ClassInfo DDoc?
 It seems, it disappeared from
 http://www.digitalmars.com/d/phobos/object.html


It's missing from the 1.010 docs (which are one the web currently), but 
reappeared in 1.011.  So it's probably in your dmd/html folder. ;)
Apr 19 2007
prev sibling next sibling parent Dan <murpsoft hotmail.com> writes:
Frank Benoit (keinfarbton) Wrote (MODIFIED BY ME)

 struct ClassInfo{
 	Field[] fields;
 	Method[] methods;
 	Constructors[] constructors;
 	static ClassInfo[] allClasses; // probably the most important
 }
 
 struct Field{
 	uint offset;
 	TypeInfo type;
 	Attribute[] attribs;
 	char[] name;
 	void* getPtr( Object* obj );
 }
 struct Parameter{
 	char[] name;
 	TypeInfo type;
 	Attribute[] attribs;
 }
 struct Method{
 	uint offset;
 	TypeInfo ret_type;
 	Attribute[] attribs;
 	char[] name;
 	Parameter[] params;
 	void* getPtr( Object* obj = null );
 }
 struct Constructor{
 	Attribute[] attribs;
 	char[] name;
 	void* getPtr();
 	Parameter[] params;
 }


That's simpler (inside)   : )
Apr 19 2007
prev sibling parent reply BLS <nanali nospam-wanadoo.fr> writes:
Still hope that you realize that Frank is going to make a zillion lines of JAVA
code available for the D community. Even if TIOPORT is far away from beeing
perfect... this tool has probabely an bigger impact as we can imagine right now.
Ergo :  Work hard <as ususal> to enable Java like reflection and forget about
the super duper D++ features for a while 
<to be continiued 
Bjoern

Walter Bright Wrote:


 Frank Benoit (keinfarbton) wrote:

 Walter Bright schrieb:

 Frank Benoit (keinfarbton) wrote:

 allClasses is already there, see Classinfo.find().


 Didn't know about it :)

 Now i can test if "org.eclipse.swt.SWT.SWT" exists.
 If so, is there a way to call the static method "static_this()"?


 In the module info, there should be an array of the static constructors.


 
 I cannot use them, because of circular dependencies. This is, why i
 "manually" call "static_this()" methods (note the underscore) in a
 defined order.


 
 There isn't a way to do that with reflection at the moment.
Apr 21 2007
parent Dan <murpsoft hotmail.com> writes:
BLS Wrote:


 Still hope that you realize that Frank is going to make a zillion lines of
JAVA code available for the D community. Even if TIOPORT is far away from
beeing perfect... this tool has probabely an bigger impact as we can imagine
right now.
 Ergo :  Work hard <as ususal> to enable Java like reflection and forget about
the super duper D++ features for a while 
 <to be continiued 


Actually, I think Reflection is going to be the biggest feature added to D
since 1.0; it'll let us easily build context sensitive code generators within
D; as we saw with that FPU vector optimization code that was in here a few days
ago.

If we get that in order, I could very readily see SSE2 and graphics
optimization code generators being written in D; which would blow C++ game
support out of the water.
Apr 22 2007