On 28/10/2016 01:22, Steve Drach wrote:
I’ve put out another webrev,
http://cr.openjdk.java.net/~sdrach/8156499/webrev.05/
<http://cr.openjdk.java.net/%7Esdrach/8156499/webrev.05/>, that
addresses Mandy’s concerns. In particular I demonstrate that the
resultant image is “runnable” and that a Main class in the image
can/cannot find the java.logging module when the module-info.class is
changed to require java.logging in one case but not in the other.
The changes to jlink look okay (same as previous round). I briefly
looked at the updated test and it looks like it does all the right
checking. Mandy had detailed comments on the tests so I'll leave that to
her.
Once this is in then the next step will be have jlink find the java.base
module and then use its version as runtime version for the module finder
that finds the modules to link into the image. The changes to jlink
should be straight-forward but testing will be complicated. I only
mention now in case it you are thinking of re-using the current test for
the second phase.
-Alan
