[rfc][icedtea-web] Signed applets with codebase loading

[Andrew Azores]

On 02/13/2014 09:50 AM, Jiri Vanek wrote:
> Hi!
>
>
> Changes to classlaoder looks ok, but test do not work:
>
> Passed: SignedAppletCodebaseLoadingTests.testCodebaseLoading - 
> chromium-browser
>  - WARNING This test is known to fail, but have passed!
> Passed: SignedAppletCodebaseLoadingTests.testCodebaseLoading - opera
>  - WARNING This test is known to fail, but have passed!
> Passed: SignedAppletCodebaseLoadingTests.testCodebaseLoading - midori
>  - WARNING This test is known to fail, but have passed!
> Passed: SignedAppletCodebaseLoadingTests.testCodebaseLoading - epiphany
>  - WARNING This test is known to fail, but have passed!
> FAILED: testSignedAppletWithExternalMainClassLaunch - 
> chromium-browser(SignedAppletExternalMainClassTest) applet did not 
> initialize
>  - This test is known to fail
> FAILED: testSignedAppletWithExternalMainClassLaunch - 
> opera(SignedAppletExternalMainClassTest) applet did not initialize
>  - This test is known to fail
> FAILED: testSignedAppletWithExternalMainClassLaunch - 
> midori(SignedAppletExternalMainClassTest) applet did not initialize
>  - This test is known to fail
> FAILED: testSignedAppletWithExternalMainClassLaunch - 
> epiphany(SignedAppletExternalMainClassTest) applet did not initialize
>  - This test is known to fail
> Total tests run: 8; From  those : 8 known to fail
> Test known to fail: passed: 4; failed: 4; ignored: 0
> Test results: passed: 4; failed: 4; ignored: 0
>
>
>
> why???
>
> (after removal of exception)
> After reading description, whole thread,and code, the second four 
> should also pass....

Well, this is incredibly embarrassing. The latest commit (mine) to head 
introduces this regression - external main-class is broken now. :(

At the time this patch was originally sent, that test would've been 
working, but it wasn't present for regression testing against that 
latest commit, so things broke...

>
> minor note:
> > +    private static void assertNoExceptions(ProcessResult pr) {
> > +        assertFalse(pr.stdout.contains("xception"));
> > +        assertFalse(pr.stderr.contains("xception"));
> > +    }
>
> Please , don't do this. I spent large amount of time in removal of all 
> those.
> nor do assertTrue(pr.stderr.contains("xception"));. Unless you are 
> searching for appearence/notappearence of some ExactException.

Okay, I'll fix this up.

>
> What about applet, which have few jars signed vith valid signature, 
> and some vith invalid???
> Isn't it also partially signed one?
> The example is recently discovered reproducer for redirection - 
> https://java.net/projects/electric/downloads/download/electric.jnlp
>
> Well it is jnlp app, but in this case it sounds to me like valid....
>
> What do you think?
>

I'm not entirely sure what you're asking here. The issues addressed by 
these patches have to do with loading un-JAR'd classes directly from the 
codebase - how is this JNLP application connected?

Thanks,

-- 
Andrew A