Re: Running StandaloneTalos test

I'm afraid that the stalling behavior is currently expected - there 
needs to be additional work done on the proxy server to see why this 
happens, but for now I would just remove those 'bad' pages from the 
manifest file.

Creating a re-playable web page test set is actually a pretty difficult 
problem.  Pages in the wild behave strangely, don't load, are full of 
errors, etc, etc.  Basically, the current best strategy is to start with 
a really large manifest and then prune it down as pages fail/don't 
load/raise errors/popups.  Eventually you'll settle on a list of pages 
that you can repeatably and consistently load and then you'll use that 
as your local test farm.

alice.

[email protected] wrote:
> On Feb 6, 3:55 pm, alice nodelman <[email protected]> wrote:
>> You aren't going to get any result from the proxy_setup.config - it
>> expressly throws away results since it is just trying to build up the
>> local cache of web pages that can be served later.  After a successful
>> run with proxy_setup.config you need to run with proxy_run.config.  You
>> can see this in the final steps on the StandaloneTalos wiki page; it has
>> you switch the proxy over to only serve pages from the local cache and
>> use the secondary config file.
>>
>> I can also see how talos is getting confused, it's expectation is that
>> the firefox path is that to the directory that contains both the
>> executable and the libraries (ie, I grab a copy of firefox fromhttp://stage.mozilla.org/pub/mozilla.org/firefox/nightly/which I then
>> untar to a local directory, then I end up with a firefox directory with
>> both firefox-bin and all the other firefox related files in it).  Having
>> your executable somewhere else from the library files meant that it was
>> unable to install the chrome/components directories.
>>
>> alice.
>>
> 
> I got the initial test framework set up and got the results following
> a run with the server using local cache. However, the test seems to
> stall now and then serving up the pages from the local cache.
> Sometimes a simple refresh gets it going and at other times I have to
> stop it and refresh it while the test is running. Is this expected
> behavior?
> 
> As you suggest, I will grab a local copy of firefox for testing and
> try this once I stabilize the framework.
> 
> thanks again,
> - mouli