[rfc][icedteaweb] Add comments about permission attributes not being checked in reproducers
Jiri Vanek
jvanek at redhat.com
Thu Mar 5 07:48:25 UTC 2015
On 03/04/2015 10:00 PM, Lukasz Dracz wrote:
> Hello,
>
> I have been looking into why certain reproducers are failing. For the following tests
>
> testPartiallySignedAppletWithSandboxPermissionsInManifestLaunchWithSignedHTMLApp
> testPartiallySignedJNLPAppletWithSandboxPermissionsInManifestLaunchWithSignedApp
> testSignedAppletWithSandboxPermissionsInManifestHtml
> testSignedAppletWithSandboxPermissionsInManifestHtmlJnlpHref
> testSignedAppletWithSandboxPermissionsInManifestJnlpApplet
> testSignedAppletWithSandboxPermissionsInManifestJnlpApplication
>
> I found that changeset 1129:0284eb954ebc is reason.
>
> The reason why is that the permissions attributes check has been moved from always being checked into an if statement
> where they are only checked when deployment.manifest.attributes.check is set to true. I have tested and found the tests to pass when this deployment.manifest.attributes.check is set to true.
>
> This patch just adds comments to the tests for reference to help explain to anyone in the future looking at the tests. At the moment I don't have a solution, does anyone have an idea on how to properly fix this or handle this ?
> I had tried to set the property to true for the tests and revert it back after to no avail. I will try to think more on this problem.
>
Hi!
So yes - the testsuite is running with deployment.manifest.attributes.check=false.
IMHO best would be to tune it to run without any deployment.manifest.attributes.check.
For your group of tests - I think the best way to go is :
@beforeclass ensure that no deployment.manifest.attributes.check is present in
deploymnet.properties. If it is, remove it.
@afterclass return it back, if it was here.
What do you think? Note - tehre is plenty util methods which loads file to string or save string to
file. Just do not add more :)
Thank you!
J.
More information about the distro-pkg-dev
mailing list