xmlvm-users — For current and interested XMLVM users, for public discussions and anouncements.

Re: [xmlvm-users] status and (L)GPL

[Arno P. <ar...@pu...>]

we did change the license to L-GPL. I haven't come around to change the 
files in the main directory of the code repository. I just did that. At 
some point we also need to change all the headers of the source files 
that still mention GPL.

I'm not sure I understand your question. The "L" in L-GPL is similar in 
nature to the Classpath Exception. You can write your own application 
with XMLVM without having to license your application under the GPL (and 
therefore keeping your application proprietary). If you make 
modifications to XMLVM, you have to contribute those back to the community.

Arno


On 12/1/10 6:36 AM, Thorsten Schemm wrote:
> Hello,
> I have a question about the projects license. From an earlier mail to
> this list I understand that Xmlvm switched to the LGPL. The SF project
> page indicates that, too.
>
> The SVN sources and a site on xmlvm.org is still mentioning the GPL.
> Maybe that should be removed at some point:
> http://xmlvm.org/download/
> http://xmlvm.svn.sourceforge.net/viewvc/xmlvm/trunk/xmlvm/
>
>
> Assuming the LGPL is correct: I was wondering if you have any tips on
> how an Xmlvm user can properly follow that license, given the
> requirement for an application user to be able to exchange the library
> part with his own version. I find this hard to achieve on the various
> mobile platforms that Xmlvm (especially with the upcoming C backend) is
> going to be able to reach. Wouldn't the Classpath Exception be more
> suitable?
>
>
> Also, would you be so kind to provide a rough estimation on when the new
> C backend will be available? I read that there were dependency problems
> if one attempted to translate parts of the OpenJDK resulting in a large
> executable.
>
> Tank you very much for your great work,
> Best regards,
> Thorsten.
>
>
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App&  Earn a Chance To Win $500!
> Tap into the largest installed PC base&  get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

[xmlvm-users] status and (L)GPL

[Thorsten S. <co...@pa...>]

Hello,
I have a question about the projects license. From an earlier mail to 
this list I understand that Xmlvm switched to the LGPL. The SF project 
page indicates that, too.

The SVN sources and a site on xmlvm.org is still mentioning the GPL. 
Maybe that should be removed at some point:
http://xmlvm.org/download/
http://xmlvm.svn.sourceforge.net/viewvc/xmlvm/trunk/xmlvm/


Assuming the LGPL is correct: I was wondering if you have any tips on 
how an Xmlvm user can properly follow that license, given the 
requirement for an application user to be able to exchange the library 
part with his own version. I find this hard to achieve on the various 
mobile platforms that Xmlvm (especially with the upcoming C backend) is 
going to be able to reach. Wouldn't the Classpath Exception be more 
suitable?


Also, would you be so kind to provide a rough estimation on when the new 
C backend will be available? I read that there were dependency problems 
if one attempted to translate parts of the OpenJDK resulting in a large 
executable.

Tank you very much for your great work,
Best regards,
Thorsten.

Re: [xmlvm-users] Network Byte Order

[Troy G. <tro...@gm...>]

Yep and going to iPhone which is little endian. 

On Dec 1, 2010, at 1:21 AM, Panayotis Katsaloulis <pan...@pa...> wrote:

> 
> On Dec 1, 2010, at 5:20 AM, Troy Gaines wrote:
> 
>> Ok, I attached a patch file.  I ended up changing readInt__, readFloat__ and readDouble__ as well.  These all compared out with my Java vs. Objective C test.
>> 
>> Thanks.  
> 
> Probably the problem was that, Java is internally always Big Endian?
> 
> 
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App & Earn a Chance To Win $500!
> Tap into the largest installed PC base & get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

Re: [xmlvm-users] Network Byte Order

[Panayotis K. <pan...@pa...>]

On Dec 1, 2010, at 5:20 AM, Troy Gaines wrote:

> Ok, I attached a patch file.  I ended up changing readInt__, readFloat__ and readDouble__ as well.  These all compared out with my Java vs. Objective C test.
> 
> Thanks.  

Probably the problem was that, Java is internally always Big Endian?

Re: [xmlvm-users] Network Byte Order

[Troy G. <tro...@gm...>]

### Eclipse Workspace Patch 1.0
#P xmlvm
Index: src/xmlvm2objc/compat-lib/objc/java_io_DataInputStream.m
===================================================================
--- src/xmlvm2objc/compat-lib/objc/java_io_DataInputStream.m	(revision 1253)
+++ src/xmlvm2objc/compat-lib/objc/java_io_DataInputStream.m	(working copy)
@@ -51,10 +51,11 @@
 	int d;
 	unsigned char* p = (unsigned char*) &d;
 	
-	for (int i = 0; i < 4; i++) {
+	for (int i = 4; i > 0; i--) {
 		int v = [target read__];
-		*p++ = (unsigned char) v;
+		p[i-1] = (unsigned char) v;
 	}
+	
 	return d;
 }
 
@@ -73,10 +74,11 @@
 	float f;
 	unsigned char* p = (unsigned char*) &f;
 	
-	for (int i = 0; i < 4; i++) {
+	for (int i = 4; i > 0; i--) {
 		int v = [target read__];
-		*p++ = (unsigned char) v;
+		p[i-1] = (unsigned char) v;
 	}
+	
 	return f;
 }
 
@@ -85,10 +87,11 @@
 	double d;
 	unsigned char* p = (unsigned char*) &d;
 	
-	for (int i = 0; i < 8; i++) {
+	for (int i = 8; i > 0; i--) {
 		int v = [target read__];
-		*p++ = (unsigned char) v;
+		p[i-1] = (unsigned char) v;
 	}
+	
 	return d;
 }
 

Re: [xmlvm-users] Network Byte Order

[Troy G. <tro...@gm...>]

I'll write a test to validate and validate.

On a related but separate note, has this project considered (if it doesn't
already have them) unit tests to validate behavior on some of these classes?
 I could see it becoming more important especially on classes that we're
implementing by hand.

Thanks again for all your help!

On Tue, Nov 30, 2010 at 12:06 PM, Arno Puder <ar...@pu...> wrote:

>
> interesting point. Can you please verify that this works now? If so, I
> will commit the patch.
>
> Arno
>
>
> On 11/30/10 7:45 AM, Troy Gaines wrote:
> > I have a char array in Objective C with this: char[] {0, 37, 0, 0}.
> >   This is later passed to the readInt__ method in
>  java_io_DataInputStream.
> >
> > This results in 9472 through the Simulator.  On the Java side, I wrote a
> > test and this results in 2424832.  It appears the byte order is being
> > reversed since  Java reads and writes everything in byte order.  It
> > seems like the following code needs to be reversed to match Java.
> >
> > So instead of p being p[0] = 0, p[1] = 25, p[2] = 0, p[3] = 0 it would
> > be p[0] = 0, p[1] = 0, p[2] = 25, p[3] = 0.  This would correctly
> > convert on the iPhone.  I posted what the new method could look like
> > although I didn't test it.
> >
> > *original*
> >
> > - (int) readInt__
> > {
> > int d;
> > unsigned char* p = (unsigned char*) &d;
> > for (int i = 0; i < 4; i++) {
> > int v = [target read__];
> > *p++ = (unsigned char) v;
> > }
> >
> > return d;
> > }
> >
> > to
> >
> > *new*
> >
> > - (int) readInt__
> > {
> > int d;
> > unsigned char* p = (unsigned char*) &d;
> > for (int i = 4; i > 0; i--) {
> > int v = [target read__];
> > p[i-1] = (unsigned char) v;
> > }
> >
> > return d;
> > }
> >
> > *
> > *
> > *
> > *
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > Increase Visibility of Your 3D Game App&  Earn a Chance To Win $500!
> > Tap into the largest installed PC base&  get more eyes on your game by
> > optimizing for Intel(R) Graphics Technology. Get started today with the
> > Intel(R) Software Partner Program. Five $500 cash prizes are up for
> grabs.
> > http://p.sf.net/sfu/intelisp-dev2dev
> >
> >
> >
> > _______________________________________________
> > xmlvm-users mailing list
> > xml...@li...
> > https://lists.sourceforge.net/lists/listinfo/xmlvm-users
>
>
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App & Earn a Chance To Win $500!
> Tap into the largest installed PC base & get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users
>

Re: [xmlvm-users] Implementing new class in Objective-C

[Panayotis K. <pan...@pa...>]

On Nov 30, 2010, at 8:18 PM, Arno Puder wrote:

> 
> the preferred way is to do as much as possible in Java and then 
> cross-compile to Objective-C. If you already have an Objective-C class 
> that you want to expose in Java, you need to create a Java stub and then 
> write some "glue" code in Objective-C that bridges the Java API to the 
> Objective-C API. This is necessary because of the certain differences 
> between the languages. E.g., we have to use name mangling when 
> cross-compiling from Java to Objective-C and the "glue" code needs to do 
> a mapping from the mangled version of a name to the original version. 

You can also take advantage of the "plugin" mechanism.
To see a demo of this convenient way, have a look at the directory
demo/iphome/plugins/Zip

Re: [xmlvm-users] Implementing new class in Objective-C

[Arno P. <ar...@pu...>]

the preferred way is to do as much as possible in Java and then 
cross-compile to Objective-C. If you already have an Objective-C class 
that you want to expose in Java, you need to create a Java stub and then 
write some "glue" code in Objective-C that bridges the Java API to the 
Objective-C API. This is necessary because of the certain differences 
between the languages. E.g., we have to use name mangling when 
cross-compiling from Java to Objective-C and the "glue" code needs to do 
a mapping from the mangled version of a name to the original version. 
Here is an example taken from 
xmlvm/src/xmlvm2objc/compat-lib/objc/org_xmlvm_iphone_UIImage.m:

+ (org_xmlvm_iphone_UIImage*) imageWithContentsOfFile___java_lang_String 
:(java_lang_String*)path
{
     return [[UIImage alloc] initWithContentsOfFile:path];
}

The new C backend (once its done), can automate some of the boilerplate 
code that currently needs to be done manually in Objective-C. Here is 
the same example from the C backend:

JAVA_OBJECT 
org_xmlvm_iphone_UIImage_imageWithContentsOfFile___java_lang_String(JAVA_OBJECT 
n1)
{
     if (!__CLASS_org_xmlvm_iphone_UIImage.classInitialized)
         __INIT_org_xmlvm_iphone_UIImage();
 
//XMLVM_BEGIN_WRAPPER[org_xmlvm_iphone_UIImage_imageWithContentsOfFile___java_lang_String]
	
     NSString *nsStr = toNSString(n1);
     UIImage *named = [UIImage imageNamed:nsStr];
	
     org_xmlvm_iphone_UIImage *toRet = __NEW_org_xmlvm_iphone_UIImage();
     toRet->org_xmlvm_iphone_UIImage.ocImage = named;
     return toRet;
     //XMLVM_END_WRAPPER
}

Only the code between the //XMLVM_* comments was written by hand; the 
rest was generated by the C backend.

Arno



On 11/30/10 1:48 AM, Tomas Vestelind wrote:
> Hi,
>
> I'm doing a proof-of-concept implementation with XMLVM for my master's
> thesis.
>
> I have a mobile application which uses a in-house framework and my goal
> is to make the program cross compile with XMLVM and do the framework
> components either in Objective-C or in Java then cross compiling it to
> Objective-C. So, I'm wondering how you do when you implement new classes
> in Objective-C. I will need this for our classes and possibly for Cocoa
> classes that are not implemented.
>
>  From what I have understood you create stubs in Java and then use XMLVM
> to generate Objective-C source files from these. Then you do the
> implementation of each class manually. Is this correct?
>
> If this is the case and I would like to implement a new class in
> Objective-C, would I do my skeleton in Java, generate the Objective-C
> code and do the implementation in Objective-C?
>
> Best regards,
> Tomas Vestelind
>
> ps. I think
> https://sites.google.com/a/xmlvm.org/documentation/home should be on
> www.xmlvm.org <http://www.xmlvm.org>
>
> --
> Quand on veut un mouton, c'est la preuve qu'on existe
>
>
>
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App&  Earn a Chance To Win $500!
> Tap into the largest installed PC base&  get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
>
>
>
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

Re: [xmlvm-users] Network Byte Order

[Arno P. <ar...@pu...>]

interesting point. Can you please verify that this works now? If so, I 
will commit the patch.

Arno


On 11/30/10 7:45 AM, Troy Gaines wrote:
> I have a char array in Objective C with this: char[] {0, 37, 0, 0}.
>   This is later passed to the readInt__ method in  java_io_DataInputStream.
>
> This results in 9472 through the Simulator.  On the Java side, I wrote a
> test and this results in 2424832.  It appears the byte order is being
> reversed since  Java reads and writes everything in byte order.  It
> seems like the following code needs to be reversed to match Java.
>
> So instead of p being p[0] = 0, p[1] = 25, p[2] = 0, p[3] = 0 it would
> be p[0] = 0, p[1] = 0, p[2] = 25, p[3] = 0.  This would correctly
> convert on the iPhone.  I posted what the new method could look like
> although I didn't test it.
>
> *original*
>
> - (int) readInt__
> {
> int d;
> unsigned char* p = (unsigned char*) &d;
> for (int i = 0; i < 4; i++) {
> int v = [target read__];
> *p++ = (unsigned char) v;
> }
>
> return d;
> }
>
> to
>
> *new*
>
> - (int) readInt__
> {
> int d;
> unsigned char* p = (unsigned char*) &d;
> for (int i = 4; i > 0; i--) {
> int v = [target read__];
> p[i-1] = (unsigned char) v;
> }
>
> return d;
> }
>
> *
> *
> *
> *
>
>
>
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App&  Earn a Chance To Win $500!
> Tap into the largest installed PC base&  get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
>
>
>
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

[xmlvm-users] Network Byte Order

[Troy G. <tro...@gm...>]

I have a char array in Objective C with this: char[] {0, 37, 0, 0}.  This is
later passed to the readInt__ method in  java_io_DataInputStream.

This results in 9472 through the Simulator.  On the Java side, I wrote a
test and this results in 2424832.  It appears the byte order is being
reversed since  Java reads and writes everything in byte order.  It seems
like the following code needs to be reversed to match Java.

So instead of p being p[0] = 0, p[1] = 25, p[2] = 0, p[3] = 0 it would be p[0]
= 0, p[1] = 0, p[2] = 25, p[3] = 0.  This would correctly convert on the
iPhone.  I posted what the new method could look like although I didn't test
it.

*original*

- (int) readInt__
{
int d;
 unsigned char* p = (unsigned char*) &d;
 for (int i = 0; i < 4; i++) {
int v = [target read__];
 *p++ = (unsigned char) v;
}

return d;
}

to

*new*

- (int) readInt__
{
int d;
unsigned char* p = (unsigned char*) &d;
 for (int i = 4; i > 0; i--) {
int v = [target read__];
 p[i-1] = (unsigned char) v;
}

return d;
}

*
*
*
*

[xmlvm-users] Implementing new class in Objective-C

[Tomas V. <tom...@gm...>]

Hi,

I'm doing a proof-of-concept implementation with XMLVM for my master's
thesis.

I have a mobile application which uses a in-house framework and my goal is
to make the program cross compile with XMLVM and do the framework components
either in Objective-C or in Java then cross compiling it to Objective-C. So,
I'm wondering how you do when you implement new classes in Objective-C. I
will need this for our classes and possibly for Cocoa classes that are not
implemented.

>From what I have understood you create stubs in Java and then use XMLVM to
generate Objective-C source files from these. Then you do the implementation
of each class manually. Is this correct?

If this is the case and I would like to implement a new class in
Objective-C, would I do my skeleton in Java, generate the Objective-C code
and do the implementation in Objective-C?

Best regards,
Tomas Vestelind

ps. I think https://sites.google.com/a/xmlvm.org/documentation/home should
be on www.xmlvm.org

-- 
Quand on veut un mouton, c'est la preuve qu'on existe

Re: [xmlvm-users] Rotate Right

[Arno P. <ar...@pu...>]

you are right. There is a cast to an unsigned int missing for the ushr 
instruction. Should be fixed now.

Arno


On 11/26/10 9:50 AM, Troy Gaines wrote:
> After further looking, it seems like the following would work for the
> corresponding unsigned shift right >>> in Java to Objective C.
>
> int k = (unsigned) k >> (0x1f & 1);
>
>
>
>
> On Fri, Nov 26, 2010 at 11:14 AM, Troy Gaines <tro...@gm...
> <mailto:tro...@gm...>> wrote:
>
>      From what I can tell, it looks like the objective C instruction
>     might not always be the same when doing a shift right.  I posted the
>     following Java code and corresponding Objective C code to illustrate
>     this.  In Java the value ends up positive and Objective C negative.
>       I'm not sure if the left shift has this same issue.
>
>     Thanks again!
>
>     int k2 = -2147483648;
>
>     k2 = k2 >>> 1;
>
>     assertEquals(1073741824, k2); //this passes
>
>
>     //objective c generated code
>
>     int_r34.i = -2147483648;
>
>     _r34.i = _r34.i >> (0x1f & 1);
>
>     STAssertEquals(1073741824, _r34.i, @"Should be 1073741824 and not
>     -1073741824");
>
>
>
>
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App&  Earn a Chance To Win $500!
> Tap into the largest installed PC base&  get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
>
>
>
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

Re: [xmlvm-users] Rotate Right

[Troy G. <tro...@gm...>]

After further looking, it seems like the following would work for the
corresponding unsigned shift right >>> in Java to Objective C.

   int k = (unsigned) k >> (0x1f & 1);




On Fri, Nov 26, 2010 at 11:14 AM, Troy Gaines <tro...@gm...> wrote:

> From what I can tell, it looks like the objective C instruction might not
> always be the same when doing a shift right.  I posted the following Java
> code and corresponding Objective C code to illustrate this.  In Java the
> value ends up positive and Objective C negative.  I'm not sure if the left
> shift has this same issue.
>
> Thanks again!
>
> int k2 = -2147483648;
>
> k2 = k2 >>> 1;
>
> assertEquals(1073741824, k2); //this passes
>
>
> //objective c generated code
>
> int _r34.i = -2147483648;
>
> _r34.i = _r34.i >> (0x1f & 1);
>
> STAssertEquals(1073741824, _r34.i, @"Should be 1073741824 and not
> -1073741824");
>
>

[xmlvm-users] Rotate Right

[Troy G. <tro...@gm...>]

>From what I can tell, it looks like the objective C instruction might not
always be the same when doing a shift right.  I posted the following Java
code and corresponding Objective C code to illustrate this.  In Java the
value ends up positive and Objective C negative.  I'm not sure if the left
shift has this same issue.

Thanks again!

   int k2 = -2147483648;

k2 = k2 >>> 1;

assertEquals(1073741824, k2); //this passes


//objective c generated code

   int _r34.i = -2147483648;

_r34.i = _r34.i >> (0x1f & 1);

   STAssertEquals(1073741824, _r34.i, @"Should be 1073741824 and not
-1073741824");

Re: [xmlvm-users] Long Type & Errors

[Arno P. <ar...@pu...>]

On 11/23/10 7:50 AM, Troy Gaines wrote:
> Ok, thanks for the quick fix!  You guys work fast.  :)

only because the issues you reported are easy to fix. Don't expect those 
turnaround times for more complicated problems. ;)

> I did see one other thing that might be a problem as well related to
> this issue.
>
> In xmlvm.m, there is this line of code which is used to allocate the
> correct array size.  Seems like it might need to be sizeof(long long).
>
>      case 8: // long
>         sizeOfBaseType = sizeof(long);

You are right. Just fixed it.

> Also, I see these all throughout my code.  I'm not sure if they're an
> issue yet.  Is there a way I can implement these?
>
> ERROR("dex:neg-long");
> ERROR("dex:shr-long-2addr");
> ERROR("dex:ushr-int-2addr");
> ERROR("dex:ushr-int");

Just added these missing instructions as well.

Arno

Re: [xmlvm-users] Long Type & Errors

[Troy G. <tro...@gm...>]

Ok, thanks for the quick fix!  You guys work fast.  :)

I did see one other thing that might be a problem as well related to this
issue.

In xmlvm.m, there is this line of code which is used to allocate the correct
array size.  Seems like it might need to be sizeof(long long).

    case 8: // long
       sizeOfBaseType = sizeof(long);

Also, I see these all throughout my code.  I'm not sure if they're an issue
yet.  Is there a way I can implement these?

ERROR("dex:neg-long");
ERROR("dex:shr-long-2addr");
ERROR("dex:ushr-int-2addr");
ERROR("dex:ushr-int");

Thanks again!


On Tue, Nov 23, 2010 at 1:28 AM, Arno Puder <ar...@pu...> wrote:

> In Java sizeof(long) == 8, so you have to map it to something
> equivalent on the target platform. For iOS this is 'long long'.
>
> Arno
>
> On Nov 22, 2010, at 10:58 PM, Panayotis Katsaloulis
> <pan...@pa...> wrote:
>
> >
> > On Nov 23, 2010, at 6:07 AM, Arno Puder wrote:
> >
> >>
> >> Wow! Now that is interesting. You are right: under iOS sizeof(long) == 4
> >> whereas sizeof(long long) == 8. I am surprised that no one ran into this
> >> problem before.
> >>
> >> Anyways, I fixed both problems in the repository. Thanks for reporting
> this!
> >>
> >> Arno
> >
> >
> > Is this the "correct" behavior for arm6/7 archs?
> > (I don't know, I just ask)
> >
> ------------------------------------------------------------------------------
> > Increase Visibility of Your 3D Game App & Earn a Chance To Win $500!
> > Tap into the largest installed PC base & get more eyes on your game by
> > optimizing for Intel(R) Graphics Technology. Get started today with the
> > Intel(R) Software Partner Program. Five $500 cash prizes are up for
> grabs.
> > http://p.sf.net/sfu/intelisp-dev2dev
> > _______________________________________________
> > xmlvm-users mailing list
> > xml...@li...
> > https://lists.sourceforge.net/lists/listinfo/xmlvm-users
>
>
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App & Earn a Chance To Win $500!
> Tap into the largest installed PC base & get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users
>

Re: [xmlvm-users] Long Type & Errors

[Arno P. <ar...@pu...>]

In Java sizeof(long) == 8, so you have to map it to something
equivalent on the target platform. For iOS this is 'long long'.

Arno

On Nov 22, 2010, at 10:58 PM, Panayotis Katsaloulis
<pan...@pa...> wrote:

>
> On Nov 23, 2010, at 6:07 AM, Arno Puder wrote:
>
>>
>> Wow! Now that is interesting. You are right: under iOS sizeof(long) == 4
>> whereas sizeof(long long) == 8. I am surprised that no one ran into this
>> problem before.
>>
>> Anyways, I fixed both problems in the repository. Thanks for reporting this!
>>
>> Arno
>
>
> Is this the "correct" behavior for arm6/7 archs?
> (I don't know, I just ask)
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App & Earn a Chance To Win $500!
> Tap into the largest installed PC base & get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

Re: [xmlvm-users] Long Type & Errors

[Panayotis K. <pan...@pa...>]

On Nov 23, 2010, at 6:07 AM, Arno Puder wrote:

> 
> Wow! Now that is interesting. You are right: under iOS sizeof(long) == 4 
> whereas sizeof(long long) == 8. I am surprised that no one ran into this 
> problem before.
> 
> Anyways, I fixed both problems in the repository. Thanks for reporting this!
> 
> Arno


Is this the "correct" behavior for arm6/7 archs?
(I don't know, I just ask)

Re: [xmlvm-users] Long Type & Errors

[Arno P. <ar...@pu...>]

On 11/22/10 7:11 PM, Troy Gaines wrote:
> Can someone tell me why I get this in the cross compiled code (Java ->
> Objective C)?
>
> ERROR("dex:add-long");

It means that there is currently no support for this instruction. I just 
added it because it was simple to fix.

> I also noticed when cross compiling some code with longs, I get overflow
> warnings in xCode.
>
> For example, I cross compile this statement in Java to Objective C:
>
> long l = 6221821273427820544;
>
> to
>
> _r13.l = 6221821273427820544;  //xCode complains about this line
>
>
> Instead of using the "long" data type in Objective C, has "long long"
> been considered.  This seems like it might fix the above issue.

Wow! Now that is interesting. You are right: under iOS sizeof(long) == 4 
whereas sizeof(long long) == 8. I am surprised that no one ran into this 
problem before.

Anyways, I fixed both problems in the repository. Thanks for reporting this!

Arno

[xmlvm-users] Long Type & Errors

[Troy G. <tro...@gm...>]

First, my compliments to the team who has worked on xmlvm.

Can someone tell me why I get this in the cross compiled code (Java ->
Objective C)?

      ERROR("dex:add-long");


I also noticed when cross compiling some code with longs, I get overflow
warnings in xCode.

For example, I cross compile this statement in Java to Objective C:

long l = 6221821273427820544;

to

_r13.l = 6221821273427820544;  //xCode complains about this line


Instead of using the "long" data type in Objective C, has "long long" been
considered.  This seems like it might fix the above issue.

Thanks for any help!

Re: [xmlvm-users] CGRect & memory issues

[Arno P. <ar...@pu...>]

this should be unrelated to CGRect. CGRect is a Java class like any 
other. This seems to be related to the reference counting mechanism of 
the Objective-C backend. When assigning a new value to myrect.size, the 
old value needs to be released first. This is something that should be 
done with the reference counting mechanism and if there is a problem I 
am really surprised because it would indicate a fundamental problem with 
all setters.

Can you mail the generated Objective-C code?

Either way, its really time to move to a proper garbage collector.

Arno


On 11/22/10 4:12 PM, Panayotis Katsaloulis wrote:
> I have a question, regarding memory issues and CGRect (and Obj-C backend).
> It seems that structs are a bit more complicated than usual.
>
>
> Let's consider the following Java code segment:
>
> CGRect myrect = new CGRect(0,0,0,0);
> // under Obj-C a CGPoint and CGSize will be allocated
>
> myrect.size = new CGSize(0,0);
> // under Obj-C a new CGSize will be applied to the myrect object,
> // but the "old" CGSize object is never deallocated
>
> if I did
> myrect.size = null;
> myrect.size = new CGSize(0,0);
> then the memory will be properly freed
>
>
> Any ideas how to get rid of this problem (apart from changing code writing style?)
> ------------------------------------------------------------------------------
> Increase Visibility of Your 3D Game App&  Earn a Chance To Win $500!
> Tap into the largest installed PC base&  get more eyes on your game by
> optimizing for Intel(R) Graphics Technology. Get started today with the
> Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
> http://p.sf.net/sfu/intelisp-dev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

[xmlvm-users] CGRect & memory issues

[Panayotis K. <pan...@pa...>]

I have a question, regarding memory issues and CGRect (and Obj-C backend).
It seems that structs are a bit more complicated than usual.


Let's consider the following Java code segment:

CGRect myrect = new CGRect(0,0,0,0);
// under Obj-C a CGPoint and CGSize will be allocated

myrect.size = new CGSize(0,0);
// under Obj-C a new CGSize will be applied to the myrect object, 
// but the "old" CGSize object is never deallocated

if I did
myrect.size = null;
myrect.size = new CGSize(0,0);
then the memory will be properly freed


Any ideas how to get rid of this problem (apart from changing code writing style?) 

Re: [xmlvm-users] synchronization of threads

[Paul P. <bay...@gm...>]

Yes, synchronization is fully supported, as well as wait, interrupt & sleep.
 Synchronization only works on objects (same in Java), so you can
synchronize on a java.lang.Boolean, but not on a boolean.

Paul

On Sun, Nov 21, 2010 at 11:46 AM, Arno Puder <ar...@pu...> wrote:

>
> Paul Poley submitted a patch for that a while ago. Yes, that should work
> (right, Paul?)
>
> Arno
>
>
> On 11/21/10 1:43 AM, Panayotis Katsaloulis wrote:
> > I remember in an older post, that there was a patch submitted that deals
> better with the synchronized methods.
> > So, can I safely assume that "synchronized" is properly supported now?
> > Can I also use for example a boolean value as a lock?
> >
> ------------------------------------------------------------------------------
> > Beautiful is writing same markup. Internet Explorer 9 supports
> > standards for HTML5, CSS3, SVG 1.1,  ECMAScript5, and DOM L2&  L3.
> > Spend less time writing and  rewriting code and more time creating great
> > experiences on the web. Be a part of the beta today
> > http://p.sf.net/sfu/msIE9-sfdev2dev
> > _______________________________________________
> > xmlvm-users mailing list
> > xml...@li...
> > https://lists.sourceforge.net/lists/listinfo/xmlvm-users
>
>
> ------------------------------------------------------------------------------
> Beautiful is writing same markup. Internet Explorer 9 supports
> standards for HTML5, CSS3, SVG 1.1,  ECMAScript5, and DOM L2 & L3.
> Spend less time writing and  rewriting code and more time creating great
> experiences on the web. Be a part of the beta today
> http://p.sf.net/sfu/msIE9-sfdev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users
>

Re: [xmlvm-users] synchronization of threads

[Arno P. <ar...@pu...>]

Paul Poley submitted a patch for that a while ago. Yes, that should work 
(right, Paul?)

Arno


On 11/21/10 1:43 AM, Panayotis Katsaloulis wrote:
> I remember in an older post, that there was a patch submitted that deals better with the synchronized methods.
> So, can I safely assume that "synchronized" is properly supported now?
> Can I also use for example a boolean value as a lock?
> ------------------------------------------------------------------------------
> Beautiful is writing same markup. Internet Explorer 9 supports
> standards for HTML5, CSS3, SVG 1.1,  ECMAScript5, and DOM L2&  L3.
> Spend less time writing and  rewriting code and more time creating great
> experiences on the web. Be a part of the beta today
> http://p.sf.net/sfu/msIE9-sfdev2dev
> _______________________________________________
> xmlvm-users mailing list
> xml...@li...
> https://lists.sourceforge.net/lists/listinfo/xmlvm-users

[xmlvm-users] synchronization of threads

[Panayotis K. <pan...@pa...>]

I remember in an older post, that there was a patch submitted that deals better with the synchronized methods.
So, can I safely assume that "synchronized" is properly supported now?
Can I also use for example a boolean value as a lock?