[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnash-dev] Re: no AVM2 for a while? (haxe tests)

From: Jonathan Crider
Subject: Re: [Gnash-dev] Re: no AVM2 for a while? (haxe tests)
Date: Mon, 10 Aug 2009 10:25:49 -0600
User-agent: Thunderbird (X11/20090608)

Benjamin Wolsey wrote:
To fix the classes.all test cases requires fixing either HAXE or our AVM2 code to execute HAXE based test cases. For more than simple test cases, I found as3compile too limited still. The tests in that directory do work well for versions older than v9, we've been using them, and I fail to see why you have a bad attitude about them. That test suite has primarily been developed by college students, give em a break.

To fix the classes.all testsuite requires going through and fixing the
results after testing with the Adobe player. It has completely nothing
to do with haxe or with our AVM2 code.

I spent quite a while running the tests with the Adobe player and posted
the results of failing tests. Some tests pass when the SWF9 version is
run and fail when lower versions are run. Some do the opposite.

What I fail to understand is why you complain to me instead of getting
it fixed. I'm sure everyone understands that having accurate results in
the testsuite is essential to avoid breaking Gnash. The fact that you've
been "using them" doesn't make them correct. On the contrary, the fact
that some tests return incorrect results means that they contribute to
getting things wrong.

The work that's been done over the summer has been partly very good from
all the interns. Anyone new to hacking on Gnash is also bound to make
mistakes. But that does not mean that it's acceptable for Gnash's
testsuite to be left in a mess. And despite my having to fix it several
times a week all summer, it still doesn't complete, precisely due to
some of those errors in the classes.all testsuite.
I will spend the rest of the time (this week) during the summer project going back and making sure all the classes.all tests work in both versions of the pp. I believe, that at the beginning we didn't have a really good understanding of haxe could do and how it worked, so that is probably why some of the test cases don't work very well. I would say that it will probably take me a day or two to get through all the cases just to see what is failing and what isn't so don't expect it all to get fixed immediately, but hopefully I'll be able to get it all done in a couple of day.

If the pp is able to execute all the test cases generated by haxe, shouldn't we expect gnash to execute them as well? It probably doesn't in some cases. It's also possible that there are some problems with haxe still because they haven't implemented everything in the same way that Adobe does in all cases. Most of the time there is a way to get haxe to generate the code that Adobe expects, but not always.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]