I was looking at bug 6208566, where the j2se build is failing on FreeBSD.

Apparently when the BOOTDIR java is run to create a java/util/CurrencyData.java
source file, hotspot is sending a warning message to stdout:
    Java HotSpot(TM) Client VM warning: Can't detect initial thread stack location
which puts it into the output file java/util/CurrencyData.java, which then
fails to compile.

Should the hotspot warning() function really be sending it's message to stdout?
Seems like it should be stderr. Should I file a hotspot bug on this, or
is there a good reason that the warning() utility in hotspot sends it's message
to stdout?

----

The more I think about it, this comes down to who owns the stdout and stderr
of a Java application. It seems like the Java application owns it, not the
VM. Most of the time, 99.99% at least, this is never a problem, but not being
able to trust that the Java application owns at least stdout, makes the use
of stdout in a Java application unstable and unreliable.

The VM is normally pretty invisible, like libc.so on Solaris. If libc.so
on solaris would occasionally sens a warning message down stdout on any
arbitrary application, I don't think people would like that much.

###@###.### 2005-1-26 16:58:14 GMT