Re: JNI bug?

[Marco Trudel <mtrudel at gmx dot ch>]

Chris Gray wrote:
> On Monday 29 January 2007 19:15, Andrew Haley wrote:
> > Andrew Haley writes:
> >  > Marco Trudel writes:
>
> >  >  > [...]
> >  >  > 	private static native void nativePrintBoolean(boolean b);
> >  >  > 	private static native void nativePrintBooleanAsInt(boolean b);
> >  >  > }
> >  >  >
> >  >  > Now this c code:
> >  >  >
> >  >  > #include <jni.h>
> >  >  >
> >  >  > JNIEXPORT void JNICALL Java_NativeBooleanTest_nativePrintBoolean
> >  >  > 	(JNIEnv *env, jclass c, jboolean b)
> >  >  > {
> >  >  > 	printf("nativePrintBoolean: %d\n", b);
> >  >  > }
> >  >  >
> >  >  > JNIEXPORT void JNICALL Java_NativeBooleanTest_nativePrintBooleanAsInt
> >  >  > 	(JNIEnv *env, jclass c, jint b)
> >  >  > {
> >  >  > 	printf("nativePrintBooleanAsInt: %d\n", b);
> >  >  > }
> >  >  >
> >  >  > [...]
>
> > Err, *why* are you doing this weird thing, anyway?  Why not pass the
> > correct type?
>
> Quite. javah or equivalent will generate a function prototype for 
> nativeBooleanPrintAsInt() with 'jboolean b', not 'jint b'. Changing the type 
> of a parameter and expecting the results to remain the same is, well, 
> excessively optimistic.

Yes, the code definitely looks wrong. But I don't know if there should 
be an automatic cast.
Anyway, it works with a Sun JRE, so we're stuck with the old discussion: 
Do we want to add support for erroneous code just to be consistent with 
a RE? There are good reasons for and good reasons against it...


Marco