On 8/14/12 1:12 PM, "labri...@digitalprimates.net"
<labri...@digitalprimates.net> wrote:
>> Given that mustella compiles swfs before running them, perhaps if we added a
>> compile-time / post-compile-time step that replaced calls to trace() with a
>> logging utility call, we could do that instead? We'd have >to re-write the
>> bit that checks the flashlog.txt to use something else, but that would also
>> mean that people wouldn't have to make a mm.cfg file on their computers
>> anymore as well. Even ErrorReportingEnable >could go away with the new
>> global exception handlers.
>>
>> Is possible to replace calls to trace() to that logging method with something
>> like as3-commons-bytecode or perhaps a tweak to mxmlc? If it's a simpler
>> fix, we could just change all the code to call that method >directly. Find
>> and replace for the win?
>
> That's the route I started going down too, however, there was some
> complication that I am just barely recall. I think it had something to do with
> either trace or other statements buried in the framework or unhandled errors.
> However, it didn't take long to find.
>
> My approach was to replace everything with a logger and then inspect that
> output. Instead of doing it a compile time though, I would just prefer to
> modify the tests. I was working on making the test runner just spin up
> multiple flash players when you ran it.
>
> Mike
Mustella already has output redirection for gathering test results. You can
configure it to use some other class to output somewhere else, even multiple
targets IIRC. It won't catch trace statements left in framework code
though.
The checkintests use flashlog.txt just because it is easier/lazier. It uses
mustella test steps, but not the harness that launches whole series of tests
and gathers and reports results. Mini_run.sh uses that harness.
--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui
