Oh ... is VF2JS still valid? I thought this approach was dropped? If it's still valid I guess there will be more work coming towars us, as I remember the two sort of having mixed up static variables ... they sort of share some parts of state ...
Oh well ... better not think of this right now ... refactoring the build to be sef-sufficient should be a great thing. Thanks for working on this. Chris -----Ursprüngliche Nachricht----- Von: Alex Harui [mailto:aha...@adobe.com] Gesendet: Freitag, 9. Oktober 2015 15:31 An: dev@flex.apache.org Betreff: Re: AW: [DISCUSS] Release Apache FlexJS 0.5.0 On 10/9/15, 12:58 AM, "Christofer Dutz" <christofer.d...@c-ware.de> wrote: >It would be great if you could split the tests up in unit-test and >integration-test. >This way the unit-tests could be executed every time and the >integration-tests once every day. > >Just give us targets to build with unit-tests but without >integration-tests and we should be fine. Makes sense. I started by looking at compiler.jx.tests. I could be wrong, but I believe all of these tests are integration tests. I don’t think they qualify as unit tests because the fire up the entire compiler. But I believe most can be converted to be “functional tests”. These tests often use spark.components.Button. I’m not sure why, maybe to test that you can load a SWC and use classes from it. However, now we have a GCL.swc in Falcon that I can use instead of spark.swc, so everything you need is in the falcon artifacts. So I am going to switch just about every test to use classes from GCL.swc. I think I am going to leave the VF2JS tests using spark and move them to a different ant target since I think VF2JS was intended to compile against spark. Thoughts? -Alex
