On Wed, 26 Feb 2020 at 20:00, Matthias Bläsing <[email protected]> wrote: > ok - I'll aks the obvious question: Why try to squeze a pointer sized > type into a int sized type? Why not use a jlong? I've been back and forth on this one - I actually implemented it both ways. I went for mapping jmethodID to jint because it didn't need any changes *at all* to the Java code or the wire protocol, and allowed all the old native libraries to work unchanged. Although more processing is needed in the native code, performance seemed OK to me, and it would be offset by not having to double the amount of data sent over the wire. For the 'all 64 bit/jlong' alternative I wasn't 100% confident I had found all places where changes were required. I also looked at a third option of having two wire protocols for 32 and 64-bit and this seemed even more complicated! Having said that, are we close to the point where all JVMs are 64 bit? Should we just get on with it and go for 64-bit? Any comments welcome! Pete ps. I look forward to discussing again in a few decades, when 128-bit pointers are the norm!!
--------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected] For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
