Testing

[Malachi de Ælfweald]

I wasn't suggesting that it was a JDK bug - but that LWJGL might be a good
test subject.

I tried the demos directly (without modification) without failure, so it is
likely my fault.  I have been trying to refactor them to be more JDK9-ish
and less redundant as a way to learn Vulkan.  Trying to build my current
codebase on JDK8 isn't currently doable (API incompatible) without some
rewrite.  Running my modification of the ColoredTriangleDemo, it sometimes
runs fine and sometimes crashes. Sometimes it runs fine, then crashes.  I'm
sure it's something I am doing since the stock demo works fine.

I need to debug through this, but have included the hs_err as requested.





Malachi de Ælfweald
http://www.google.com/profiles/malachid

On Sun, Jul 24, 2016 at 10:06 AM, Philip Race <philip.race at oracle.com>
wrote:

> The odds that this is jigsaw related are zero.
> The odds that this is a bug in JDK 9 are only slightly above that.
>
> LWJGL apps crash all the time due to driver issues completely outside the
> JDK.
>
> And the entire Vulkan stack and APIs are drivers are all EA so far as I
> know.
> Hmm .. the internet tells me the first ever stable release of the vulkan
> *specification* was exactly 2 weeks ago. Its that new.
> Gobs of low-level native code and I expect the JDK is not implicated
> in any way.
>
> -phil.
>
>
> On 7/24/16, 1:31 AM, Malachi de Ælfweald wrote:
>
>> I'll try to validate against 8 and see. I did notice it was mostly in
>> jemalloc and pthread.
>>
>> On Jul 24, 2016 1:15 AM, "Alan Bateman"<Alan.Bateman at oracle.com>  wrote:
>>
>> On 24/07/2016 04:52, Malachi de Ælfweald wrote:
>>>
>>> If we are looking for people to validate the jigsaw builds, we may want
>>> to
>>>
>>>> reach out to LWJGL.  I've noticed quite a few (but inconsistent) SIGSEV
>>>> when using Jigsaw on their Vulkan demos:
>>>>
>>>>
>>>> https://github.com/LWJGL/lwjgl3-demos/tree/master/src/org/lwjgl/demo/vulkan
>>>>
>>>> ​Anyone want to validate that it isn't just me?​
>>>>
>>>> LWJGL issues can sometimes be OpenGL but it's impossible to say here. If
>>>>
>>> it looks like this is a JDK 9 bug (you don't see it with JDK 8 for
>>> example)
>>> then can you submit a bug on this? Ideally include the error log (hs_err*
>>> file) if you can find it.
>>>
>>> -Alan
>>>
>>>