Re: Adding WebSocket and WebWorker test suites to testing framework

On Fri, 8 Mar 2013, Ken Reed wrote:
> Thanks Simon.
> 
>  
> 
> Did these updates potentially impact the earlier set which James submitted Feb 28th? I’m still getting timeouts on the binary/
> tests but now 2 and 4 actually hang for quite a while before returning with a Timeout. Previously they returned relatively
> quickly with a Timeout.

Simon fixed a few bugs in the set that I thought were ready; in those 
cases he upped the timeout because the amount of data being trasferred is 
quite large.

> Regarding the Timeout errors for the binary/tests. Based on Mike’s suspicion that Microsoft addresses might be getting some sort
> of special treatment from the W3C network I tried a couple other environments. Taking the same laptop which was failing via
> Microsoft's network I attempted to run the variations by connecting via a Clear Wire modem instead. The test continued to suffer
> timeouts. I then took the laptop home and tried via my Comcast connection and the test actually passed! I only ran them a few
> times each so I don't know that's a sufficient sample to conclude they would pass consistently. Instead, I proceeded to run
> through the remaining tests but started hitting errors in the constructor/ set which reproduced on both IE and Opera. Per James's
> mail I understood he expected all of these test cases were passing. I had not made it this far down the list previously so don't
> know if something regressed or this might also be related to the other failures.

I think Simon fixed some of these errors.

> James, certainly debugging these issues would be a challenge if you don't have a local repro. Is there any sort of tracing I
> could collect for you to help you debug these issues? It's quite difficult for us comment on the appropriateness of including
> these in the approved set if we can't run them.

Yes, I understand. I think we need to determine whether the problem is on 
the W3C side or on the Microsoft side. I don't really think that I can 
help with that since I can't see the logs. I wonder if you tried the tests 
with ?wss appended to the urls to get the secure versions. That would at 
least give an indication of whether the issue was something specific to 
port 80.