This patch fixes issue in the way that it:
a) calculates the result before call to (*env)->ReleaseStringUTFChars
b) attempt to trick gcc optimizations by complicating path

I suppose issue is with what (*env)->ReleaseStringUTFChars does and
how.

One should note that similar approach is not applicable to substitute
in package definition.

Why not just build ancient code with ancient gcc?



Reply via email to