On Wed, 2004-04-28 at 12:57, Tom Tromey wrote: > >>>>> "Michael" == Michael Koch <[EMAIL PROTECTED]> writes: > > Michael> I found out that in jni.h from GCJ (3.3 and 3.5) and jni.h > Michael> from Mozilla GetStringUTFChars returns "const char *" which is much > Michael> more logical and seems to be the right thing. JNI 1.1 spec says it > Michael> should return "const jbyte *". I search a little bit with google and > Michael> jni.h in other open source projects use "const char *" too. > > IMO it should be `const char *'. My reasons are that first, Sun does > this, so presumably most existing JNI source relies on this; second, > this makes more sense; and third, this is what the analogous > GetStringUTFRegion uses. > > Note that your patch is insufficient. You must also update the > GetStringUTFChars call in the C++ class later on in the file. > > I think classpath and libgcj should completely share jni.h. This > probably isn't too hard to get to, since the current classpath version > is a small fork from some earlier libgcj version. We could put the > gcj-specific stuff into a `jni_md.h' that is gcj-specific if that > would make the task easier. > > Also, why is the file `jni.h.in'? I didn't see any configure > substitutions in it.
I think because jni_md.h is generated and to simplify the INCLUDES in Makefile.am during build. Brian
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Classpath mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/classpath
