Hi,

congrats Greg for get this one complete! I shared this in twitter, Linkedin
and facebook Royale social media accounts:

https://twitter.com/ApacheRoyale/status/1138438001971871744

@all: please share so we can spread the word! :)

Best

Carlos



El mar., 11 jun. 2019 a las 8:24, Greg Dove (<[email protected]>)
escribió:

> Hi all,
>
> Just a quick summary of things merged in via recent commit.  See below for
> "How do I get things 'how they were' ?" if you want to switch off any of
> the new stuff.
>
> More reliable actionscript support.
> The new settings provide greater compatibility. But they can also be tuned
> out (Hello world is a good example, and it is smaller now in gzipped format
> for release build, than it was before)
> -More things will 'just work' when porting from older actionscript libs.
> Vector behaviour is virtually identical to swf, other types of implicit
> type coercions are generated by default (but avoidable by config), and
> certain uses of int, uint and Class are more reliable and avoid compiler or
> runtime errors compared to before.
> Reflection now has PAYG/opt-in support for including top level as3 types
> (Array, Number, int, Vector etc). These roughly match the results you see
> with reflection in swf (actually reflection info for these swf items quite
> light). This is via a plugin style setup that is only included in release
> build if you use it. See here:
>
> https://github.com/apache/royale-asjs/blob/c6379b85d8aa9d164d49b09402a23f30cb4aee3a/frameworks/projects/Reflection/src/main/royale/org/apache/royale/reflection/ExtraData.as#L127
> This same approach could be used with externs reflection data if it could
> be generated in a similar format, e.g. to provide possible reflection pack
> support for sets of javascript native classes.
> I plan to use the same ExtraData class to add support for Vector to
> AMFBinaryData, Vector is a low level type that has its own serialization
> rules. In this way there will also be no hard dependency on vector support
> in AMFBinaryData and it is therefore opt-in/PAYG via the ExtraData class.
>
> How do I get things 'how they were' ?
> maven (via additionalCompilerOptions):
>             -js-complex-implicit-coercions=false;
>             -js-vector-index-checks=false;
>             -js-resolve-uncertain=false;
>             -js-vector-emulation-class=Array;
> (note that you should also now be able to have these on separate lines in
> the pom - I find this much easier to read)
>
> ant:
> in compile-js-config.xml example
>     [js-complex-implicit-coercions]false[/js-complex-implicit-coercions]
>     [js-resolve-uncertain]false[/js-resolve-uncertain]
>     [js-vector-index-checks]false[/js-vector-index-checks]
>     [js-vector-emulation-class]Array[/js-vector-emulation-class]
> (angle brackets are substituted here with square brackets above in case
> they mess with some email clients)
>
> The first 3 options above have doc comment directives (@royalesuppress... )
> so they can be tuned off or on inside method definitions also.
>
> Fixes
> Vector-as-Array emulation now supports insertAt/removeAt (same as Array)
> vectorEmulationClass receives the fully qualified name as the string
> parameter for the elementType
>
>
> Other
> -Added a lot of compiler tests for the vector emulation options and default
> vector implementation
> -RoyaleUnit: I ported a start of the manualtest unit tests to RoyaleUnit.
> @josh... these are actually language tests, but I thought they could not
> run in the Language project because it does not have all dependencies to
> run the tests themselves, perhaps I am wrong about that? I put them in Core
> for now, and they do run correctly in both swf and JS. But please tell me
> what you prefer for these (maybe they do need to go in Language...). Also I
> made some small changes in the js ant scripts for tests to correctly clean
> the test folder.
> -Added more reflection unit tests (manualtests)
> -Started to collate unit tests for XML and updated code to fix some
> observed issues (compiler and framework).
> -Hello world is approx 600 bytes smaller after gzip compression than it was
> before
>
> I might have forgotten some things, I will add extra info tomorrow if I
> recall more. Please ask questions if you have them.
>


-- 
Carlos Rovira
http://about.me/carlosrovira

Reply via email to