[icedtea-web][rfc] Fix to AppletBaseURL reproducer

[Jiri Vanek]

On 10/16/2012 10:30 PM, Omair Majid wrote:
> On 10/16/2012 04:24 PM, Adam Domurad wrote:
>> I would argue that the 'xception' matcher is overly broad and a test
>> weakness. Currently there is a NPE while looking up the security context
>> - which I think is caused by this exit() call.
>
> I am surprised. This test is running inside a browser right? Could you
> take a look into why NPE is happening when you have some time?
>
>> However, there's a lot of
>> things that can cause exceptions on shutdown, I think (Jiri may know
>> better here.) This is why I'm in favour of changing 'xception' to the
>> appropriate exception, or getting rid of it if its just a catch-all.
>
> I don't really like the idea of 'appropriate exception'. If an exception
> happens (and not just looged to stdout), things are not working as
> desired and the test should fail. What is an appropriate exception? Who
> determines if an exception is appropriate?
>
> Does it really matter what exception is causing things to not function?
>
> That said, I am aware of errors like ThreadDeath causing exceptions in
> random places (such as in locking code) that really cant be worked
> around in icedtea-web.
>
>> Updated patch attached, with 'xception' matcher still removed.
>
> Looks fine to me, though it includes unrelated changes from another patch.
>

I had similar feelings as Omair when I was writing *xception. However it really is appearing as bad idea. The main reason is occasional "random" exception thrown from browser. I assume this is happening because of killing it so many times. I'm looking forward whamagic will do Pavels "window closer".

Thank you both for elaborating on this issue.

J.