[rfc][icedtea-web] PR1592 reproducer fix

[Jiri Vanek]

On 12/04/2013 10:17 PM, Andrew Azores wrote:
> Hi,
>
> Jiri pointed out that the JNLP versions of the tests don't close properly on their own. This patch makes them well behaved :)
>
> Fix for JNLP tests in PR1592 reproducers not closing properly
> * tests/reproducers/signed/MixedSigningAppletSigned/srcs/MixedSigningAppletSigned.java: handle JNLP "argument" tags
> * tests/reproducers/simple/MixedSigningApplet/resources/MixedSigningApplet-1.jnlp: change <param> to <argument>
> * tests/reproducers/simple/MixedSigningApplet/resources/MixedSigningApplet-2.jnlp: same
> * tests/reproducers/simple/MixedSigningApplet/resources/MixedSigningApplet-3.jnlp: same
> * tests/reproducers/simple/MixedSigningApplet/resources/MixedSigningApplet-4.jnlp: same
> * tests/reproducers/simple/MixedSigningApplet/resources/MixedSigningApplet-5.jnlp: same
> * tests/reproducers/simple/MixedSigningApplet/resources/MixedSigningApplet-6.jnlp: same
>
> Thanks,
>

This do not seem to be correct.


Please revise your approach.  Each applet can be both appelt and jnlp app with main and vice versa.

BY this you submited yourself to greater task.

The reproducers must work :
  as applet in html
  as jnlp application
  as applet lunched via jnlp

This will increase the number of jnlp files and of reproducer's testcase test methods, but *not* the sources of reproducers. Standart aproach is to have body of application called from start or main.

This is also pitfall of both your fix(jnlp part) and reproducer. You were never testing jnlp application which was really requesting any permissions. The security tag was missing for all your application desc (my overlook during firts review)


J.