> I uploaded a pretty big patch with various ObjC - Java API improvements.
Yay, great! I love your patch set already!
I think xmlvm has great potential and it works very surprisingly nicely already.
It's good that you made these patches public now, as I think most
people starting to look into xmlvm to generate Objective-C from Java
bytecode and will right away notice the same deficiencies as you did,
and attempt fix them. (Well, at least I did... but not to the extent
you have done.)
Duplicating work is a waste of effort, and noticing one has wasted
time on something that others have already solved in a better way
makes a potential contributor frustrated...
> We are prepared to rebase it against the latest trunk, and also to split it
> up to smaller, self contained changes during the review process.
Although I obviously can't speak for the xmlvm developers I think that
is what one would expect, yes.
I am just a newbie to xmlvm, just found it last week, and I am
considering using it for some small scale spare time hacking
("hacking" in the non-evil sense) for the iPhone. (Not sure what if
anything will come out of it and how much time I will have on it.)
Anyway, looking at your patch set, I see several pointless diff chunks
that just add or remove comments, like this:
* For more information, visit the XMLVM Home Page at http://www.xmlvm.org
*/
+/**
+ *
+ *
+ */
+
#import "xmlvm.h"
Probably these should be dropped to make a review easier. Are these
comments supposed to have some semantic meaning, as I see them added
in several places.
Here BTW is a patch for something I think you don't cover: native methods.
Index: xmlvm/src/xmlvm2objc/xsl/xmlvm2objc.xsl
===================================================================
--- xmlvm/src/xmlvm2objc/xsl/xmlvm2objc.xsl (revision 940)
+++ xmlvm/src/xmlvm2objc/xsl/xmlvm2objc.xsl (working copy)
@@ -374,10 +374,55 @@
</xsl:when>
<xsl:when test="@isNative = 'true'">
<xsl:text>{
- NSException* ex = [[NSException alloc] initWithName:@"Native
method not implemented" reason:nil userInfo:nil];
- @throw ex;
-}
+ extern </xsl:text>
+ <xsl:call-template name="emitType">
+ <xsl:with-param name="type"
select="vm:signature/vm:return/@type"/>
+ </xsl:call-template>
+ <xsl:text> </xsl:text>
+ <xsl:value-of select="vm:fixname(../@package)"/>
+ <xsl:text>_</xsl:text>
+ <xsl:value-of select="../@name"/>
+ <xsl:text>_</xsl:text>
+ <xsl:call-template name="emitMethodName">
+ <xsl:with-param name="name" select="@name"/>
+ <xsl:with-param name="class-type"
select="concat(../@package, '.', ../@name)"/>
+ </xsl:call-template>
+ <xsl:call-template name="appendSignature"/>
+ <xsl:text> (</xsl:text>
+ <xsl:for-each select="vm:signature/vm:parameter">
+ <xsl:call-template name="emitType">
+ <xsl:with-param name="type" select="@type"/>
+ </xsl:call-template>
+ <xsl:if test="position() < count(../vm:parameter)">
+ <xsl:text>, </xsl:text>
+ </xsl:if>
+ </xsl:for-each>
+ <xsl:text>);
+ </xsl:text>
+ <xsl:if test="vm:signature/vm:return/@type != 'void'">
+ <xsl:text>return </xsl:text>
+ </xsl:if>
+ <xsl:value-of select="vm:fixname(../@package)"/>
+ <xsl:text>_</xsl:text>
+ <xsl:value-of select="../@name"/>
+ <xsl:text>_</xsl:text>
+ <xsl:call-template name="emitMethodName">
+ <xsl:with-param name="name" select="@name"/>
+ <xsl:with-param name="class-type"
select="concat(../@package, '.', ../@name)"/>
+ </xsl:call-template>
+ <xsl:call-template name="appendSignature"/>
+ <xsl:text> (</xsl:text>
+ <xsl:for-each select="vm:signature/vm:parameter">
+ <xsl:text>n</xsl:text>
+ <xsl:value-of select="position()"/>
+ <xsl:if test="position() < count(../vm:parameter)">
+ <xsl:text>, </xsl:text>
+ </xsl:if>
+ </xsl:for-each>
+ <xsl:text>);
+ }
+
</xsl:text>
</xsl:when>
<xsl:otherwise>
With this patch, native method definitions in a Java class like:
private static native int foo1(int a, float b);
private static native void foo2();
get translated into:
+ (int) foo1___int_float :(int)n1 :(float)n2
{
extern int fi_iki_tml_test1_foo1___int_float (int, float);
return fi_iki_tml_test1_foo1___int_float (n1, n2);
}
+ (void) foo2__
{
extern void fi_iki_tml_test1_foo2__ ();
fi_iki_tml_test1_foo2__ ();
}
I.e. a native method turns into an Objective-C stub that calls the
actual extern function with "normal" C linkage. I think supporting
native methods is essential to make it possible to actually use Java
code for JDK compatibility classes (possibly even suitably licensed
existing code), with just the minimal amount of code in Objective-C or
plain C.
--tml
|
