Re: RFR: 8171400: Move checking of duplicate packages in the boot layer to link time

[Jonathan Gibbons]

Looks OK to me.

-- Jon

On 12/19/2016 02:41 PM, Mandy Chung wrote:

tools/launcher/modules/patch/systemmodules/PatchSystemModules.java needs to be 
updated since ModuleBootstrap now depends on this new method:

diff --git 
a/test/tools/launcher/modules/patch/systemmodules/src1/java.base/jdk/internal/modules/SystemModules.java
 
b/test/tools/launcher/modules/patch/systemmodules/src1/java.base/jdk/internal/modules/SystemModules.java
--- 
a/test/tools/launcher/modules/patch/systemmodules/src1/java.base/jdk/internal/modules/SystemModules.java
+++ 
b/test/tools/launcher/modules/patch/systemmodules/src1/java.base/jdk/internal/modules/SystemModules.java
@@ -29,4 +29,8 @@
   */
  public final class SystemModules {
      public static final String[] MODULE_NAMES = new String[0];
+
+    public static boolean hasSplitPackages() {
+        return true;
  }
+}

Since this fix has been pushed, I will fix this with a separate issue.

Mandy

On Dec 19, 2016, at 4:30 AM, Claes Redestad <claes.redes...@oracle.com> wrote:

Hi,

this patch adds a check to see if there are any split packages in the system
modules at link time, and uses this information to enable us to safely skip
a runtime check during bootstrap for the common case that there are none
of the sort.

Webrev[1]: http://cr.openjdk.java.net/~redestad/8171400/webrev.01/
Bug: https://bugs.openjdk.java.net/browse/JDK-8171400

This removes a chunk of the module system bootstrap overhead, and also
amends a small issue where PACKAGES_IN_BOOT_LAYER would be wrong in the
presence of split packages.

Thanks!

/Claes

[1] Since cr.openjdk.java.net is down I've also attached the raw patch.
<jdk.changeset>