XTSocketReadingWritingTest>>#testReadWriteLargeAmount fails

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

XTSocketReadingWritingTest>>#testReadWriteLargeAmount fails

Frank Shearar-3
Hi,

I loaded ConfigurationOfXtreams into Trunk, and I have the above test
as a single failure.

It's not obvious to me why the test should fail, and I think I
remember it passing in the past (because I'm pretty sure I'd have
moaned about the failure already!). It's not as simple as not having
enough time to run; upping the timeout to 10s makes no difference. And
yet the test does _occasionally_ pass.

Any ideas?

frank

Reply | Threaded
Open this post in threaded view
|

Re: XTSocketReadingWritingTest>>#testReadWriteLargeAmount fails

Colin Putney-3
On Fri, Jun 15, 2012 at 10:07 AM, Frank Shearar <[hidden email]> wrote:

> Hi,
>
> I loaded ConfigurationOfXtreams into Trunk, and I have the above test
> as a single failure.
>
> It's not obvious to me why the test should fail, and I think I
> remember it passing in the past (because I'm pretty sure I'd have
> moaned about the failure already!). It's not as simple as not having
> enough time to run; upping the timeout to 10s makes no difference. And
> yet the test does _occasionally_ pass.
>
> Any ideas?

I think this is a flaw in the Xtreams tests. They're actually doing
socket IO, which means spawning processes. I think it's an
interprocess timing issue. Multi-process tests are just about
impossible to do well.

I don't know what the right fix for this is. In the past I've used a
Mocket to test socket IO, but I don't know if that's appropriate here.

Colin