The current annotation scheme is based on some toplevel annotations (e.g. @NativeHeader) which declare native function descriptors and global variable layouts. These descriptors/layouts then contain layout annotations to point 'forward' to the Java member that references them.

This scheme blows up the Utf8 size limit for complex libraries, as the entire library is lowered onto a single string. A better approach is to split the monolithic header into multiple member annotations, as proposed here:

http://cr.openjdk.java.net/~mcimadamore/panama/panama-annos.html