Andy,
I'd propose instead of checking if the app is bundled with proper
runtime in the test case, enforce the test case to always create runtime
for the app with JPackageCommand.ignoreDefaultRuntime(true) call:
---
JPackageCommand cmd = JPackageCommand.helloAppImage(
71 JavaAppDesc.parse("com.hello/com.hello.Hello"))
...
76 .addArguments(...)
.ignoreDefaultRuntime(true);
---
With ignoreDefaultRuntime(true) call added there is no need to call
JPackageCommand.isFakeRuntime().
Is com.hello/com/hello/Hello.java used at all? I'd guess after we had
refactor all package tests this source is not used any more. Can it be
removed yet?
- Alexey
On 8/24/2020 4:50 PM, Andy Herrick wrote:
please review revised webrev at:
http://cr.openjdk.java.net/~herrick/8251988/webrev.03/
jpackage.app-version set in cfg file, and jpackage.app-path now set
from native code.
New testcase in BasicTest to check that they are set.
/Andy
On 8/20/2020 7:36 PM, Alexey Semenyuk wrote:
Andy,
I'd replace 'File.separator' with '/' to make value of
'jpackage.app-path' property platform independent.
It would be also good to have a unit test verifying values of the new
properties.
- Alexey
On 8/20/2020 5:44 PM, Andy Herrick wrote:
Please review the jpackage fix at [1] to issue [2].
This fixes the problem loading libjli.dylib as well as providing
some useful generic system properties.
/Andy
[1] - http://cr.openjdk.java.net/~herrick/8251988/webrev.02/
<http://cr.openjdk.java.net/~herrick/8251988/webrev.01/>
[2] - https://bugs.openjdk.java.net/browse/JDK-8251988
