On Tue, Mar 11, 2014 at 11:15 PM, Alex Harui <aha...@adobe.com> wrote:
> > > On 3/11/14 10:26 PM, "OmPrakash Muppirala" <bigosma...@gmail.com> wrote: > > >Doesn't Mustella need bitmap comparison? That would be almost impossible > >to get to work in HTML. Selenium probably is a better option for these > >kind of things. > Well, lately I've been wishing we'd done display-list compares instead of > bitmap compares. Just think of how much effort goes into maintaining the > bitmaps, how many are wrong, etc. So it isn't "need". We're going to > have to touch virtually every existing test. I don't think we can get > them to run as-is. The question is how much we need to touch. > I think bitmap comparisons are the most fool-proof way way to test things, but that is just me. If we don't want to do bitmap comparison, perhaps we can use FlexUnit? > > > >Another option is that perhaps we could build a runner in AIR that loads > >the html stuff in a HTML component. That we could reuse the bitmap > >comparison part as well. That would test only Webkit rendering, but that > >could be a good baseline. > That's a cool idea. We'll call that #6. > There is also the offer that Tigran from RiaTest made earlier. This thread has the details [1] He even created a sample test here [2] > > > >I was thinking of something like this for when we need to start testing > >the > >FXG/SVG skins. > Unlike flash.display.Graphics, where you find out what commands were > called, can you get back the SVG tree? > Everything is in the DOM, so walking through it and checking the properties should work, right? [1] http://markmail.org/message/khghskwyz4okyph3 [2] https://issues.apache.org/jira/browse/FLEX-33489 > >>I was looking at the Marmotinni stuff that Erik did a while back that > >>uses > >> Selenium to run tests in the browsers. The actual test seem to be > >>written > >> in Java. I think it would be nice to be able to repurpose the MXML > >> mustella tests and get them to run as cross-compiled tests. I think > >>some > >> choices are: > >> > >> 1) See what happens when Mustella is cross-compiled. I'm pretty sure > >>this > >> won't work. It could be an early test of what it will be like to > >>support > >> other third-party AS frameworks, but Mustella is relying on some > >>low-level > >> Flash things like frame events, security sandboxes, etc, so I don't > >>think > >> it is a fair test. We could fork Mustella and strip some of that stuff > >> out though, but I'd rather not have a fork of Mustella around to > >>maintain. > >> 2) Make parallel Mustella JS files. That's the direction I'm taking > >>right > >> now. That's a valid FlexJS way since FlexJS is mostly about parallel > >> AS/JS frameworks. I still need a way to run the test in the browser and > >> collect output. Hence the desire to see if Marmotinni could do that. > >> 3) Write some test converter that converts a Mustella MXML script into > >> Java code or maybe an XML representation that some Java code could > >> interpret to execute the test. > >> 4) Just re-write every test in Java. > >> 5) Something I haven't thought of yet. > >> > >> Thoughts? > >> -Alex > >> > >> > >
