|| CSS + Inline script test results misleading

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Run Network tests on Chrome 5.
2. Notice it fails "|| CSS + Inline script" test even though Chrome doesn't 
block 
in this case.

What is the expected output? What do you see instead?
Here are the possible behaviors:
1. Block download and block script execution
2. Parallelize download but block script execution
3. Parallelize download and parallelize execution

Browserscope wants #2. But lumps #1 and #3 into the same failure case even 
though 
#1 is a performance issue and #3 is a correctness issue. Since all the other 
Network tests only check performance issues, it is misleading when "|| CSS + 
Inline scripts" fails for a correctness issue rather than a performance issue. 
The ordering of script execution is not really a Network test, it is a renderer 
test. I'd like it to indicate that in some way.

Two potential solutions:
1. Split into two tests, one that checks for #1 and one that checks for #3.
2. Change the result from a boolean No/Yes to a tri-state (No/Yes/Yes* -- where 
"Yes*" explains the correctness issue).

Original issue reported on code.google.com by tonyg@chromium.org on 22 Apr 2010 at 11:54

[GoogleCodeExporter]

I would like the test to show three states. I had in mind the following:

   No: Blocks
   No*: Issue with Correctness
   Yes: Non-blocking

To mean, correctness should not be sacrificed for performance.

Original comment by steve.lamm on 27 Apr 2010 at 7:02

[GoogleCodeExporter]

I still don't see how testing for order of execution is a network test. But as 
long as there's some indication of the 
tri-state, I'm happy.

Original comment by tonyg@google.com on 27 Apr 2010 at 11:36

[GoogleCodeExporter]

We don't want browsers to implement performance improvements in a way that 
breaks
correctness. It's not true that all the other Network tests only check 
performance
issues. The "Async Attribute for Scripts" and "Parallel Stylesheet and Inline 
Script"
tests also have correctness checks. I agree it's a different type of failure, 
but to
me, it's a failure nonetheless.

If we all agree this is an issue, then it'd be better if Chrome fixed the 
issue. 

Original comment by stevesou...@gmail.com on 29 Apr 2010 at 5:11