If you're interested, the final version is on:
http://cr.openjdk.java.net/~wetmore/7197071/webrev.02/
I removed the UnsatisfiedLinkError check, and also added a check for
jce.jar, and did some minor housecleaning.
Putback pending a build from RE.
Thanks,
Brad
On 9/12/2012 12:54 PM, Valerie (Yu-Ching) Peng wrote:
The enhanced test is run for all platforms, right?
However, the particular test for the PKCS11 using Solaris crypto library
is platform specific. So, it seems kind of strange to have them in the
same test which runs on all platforms.
In addition, I don't see a need for testing this UnsatisfiedLinkError
actually. You should never encounter this error when running against
Solaris crypto impl since they define these PKCS11 entry points. Why
testing something always pass?
Thanks,
Valerie
On 09/12/12 11:41, Sean Mullan wrote:
Looks good to me.
--Sean
On 09/12/2012 03:39 AM, Brad Wetmore wrote:
Valerie/Sean,
Here's the change for JDK8.
7197071: Makefiles for various security providers aren't including the
default manifest.
I've added the standard/regular rt.jar attributes (see
common/Release.gmk) to the remainder of the JCE jars (see
javax/crypto/Makefile and com/sun/crypto/provider/Makefile), and my
regression test makes sure that the Specification version value lines up
with the Implementation Version value.
This does have the side effect that when we are transitioning major
releases, this test will fail until the providers are rebuilt. This is
an acceptable amount of noise (once per release), since that should is
an actual failure.
One minor addition from the previous version you saw, the reg test has
been enhanced, and the Defs-jce.gmk file was updated to remove the -ea
string, which was causing some confusion.
Thanks,
http://cr.openjdk.java.net/~wetmore/7197071/webrev.01/
Brad
