I've fixed the http module tests. The problem was that one of the
tests didn't shutdown a server connection properly, so following tests
fail. In JDK5 that test was always run as first, in JDK6 - among last,
that's why we saw the issue on JDK 5.
For sure this is true, if I saw the same issue as you did.
WBR,
Alexey.
On Apr 16, 2010, at 10:48 , Oleksiy Stashok wrote:
>> OK. So i have some questions for everyone. We dont have to answer
>> them now, but perhaps at next week's tech meeting if us Oracle
>> types are not tied up in planning meetings next week.
>> Given that the tests don't run successfully on 1.5 for some strange
>> reason, do we want to maintain 1.5 compatibility for the 1.9
>> branch? I'm currently running a release on a 1.6 VM and targetting
>> the 1.5 byte code. While this generates (hopefully) 1.5 compliant
>> byte code, it doesn't really catch when we use 1.6 only classes
>> like I did in websockets. If everyone still needs 1.5 compatbility
>> I guess we can hobble along like this. I know we talked about that
>> a while back but it's still a burr in my saddle so to speak that we
>> have to jump through hoops like this. If we stay with 1.5, we need
>> to figure out why the tests crash on a 1.5 VM.
> I'd stay with 1.5, as it was planned time ago. I'll take a look at
> failures and try to fix them asap.
> BTW I see tests pass fine with 1.9.18-x, but fail with 1.9.19 trunk,
> is it what you're seeing?
>
>> Does anyone have any objections to me migrating our tests from
>> junit to testng? I much prefer testng and we get quite a bit more
>> control and features for very little effort. This is probably the
>> least contentious of the two but I wanted to throw that out there.
> I'm ok with testng migration. AFAIR we use it with grizzly-config
> and nobody had problems with them. As for Netbeans - I didn't fill
> any diff. comparing to junit.
>
> WBR,
> Alexey.