Now, strangely, in both Windows CMD or WSL, I see the same RAT error when trying to build with maven.
I have verified that my git clone is the latest and clean. From: Brian Raymes <[email protected]> Sent: Tuesday, March 10, 2020 4:25 PM To: [email protected] Subject: RE: VSCode + Royale SDK: Ant != Maven Distribution Currently building this on the Windows command line as it fails in WSL Debian (see below). Windows: Source location: c:\dev\royale-asjs Maven build (tried with and without the swf-option): mvn clean install -P with-examples,with-distribution,option-with-swf -DdistributionTargetFolder=c:\dev\royale-sdk When I point vs-code to c:\dev\royale-sdk for the SDK, all of the errors and issues I listed before, are seen. If I simply perform `ant all` in c:\dev\royale-asjs, and point vs-code to c:\dev\royale-asjs, all is well. This workaround is fine for me for now, but could be problematic for others. WSL Debian: I wish I could build it here as most of my dev work is done within WSL, but, unfortunately, it fails every time with this license issue: [ERROR] Failed to execute goal org.apache.rat:apache-rat-plugin:0.13:check (license-check) on project royale-framework-parent: Too many files with unapproved license: 1 See RAT report in: /mnt/c/dev/royale-asjs/target/rat.txt -> [Help 1] The license in question is: !????? js/lib/dom4j-LICENSE.txt ----------------------- I can start creating issues for anything that I find on the github project if preferred. Brian From: Carlos Rovira <[email protected]<mailto:[email protected]>> Sent: Tuesday, March 10, 2020 3:46 PM To: [email protected]<mailto:[email protected]> Subject: Re: VSCode + Royale SDK: Ant != Maven Distribution Hi Brian, Maven should give you code completion without problem. The only issue right now with Maven SDK is if you want to build using SDK in an IDE (in VSCode CMD+SHIFT+ENTER), instead to build with normal maven command (mvn install) using a pom.xml (as normal). Building with IDE was working some weeks ago for VSCode, but Moonshine reported it was not working and I couldn't get back to fix that issue. But as I said, I think is not critical and should not be a problem in your case. * Are you generating the SDK with Maven with: -P with-distribution ? * Where are you generating the SDK? by default is in royale-asjs/distribution/target, but you can generate in other folder using -DdistributionTargetFolder=/path-to-folder think that ANT creates SDK in royale-asjs folder directly, while maven is not doing that, so you can have ROYALE_HOME pointing to royale-asjs, for ANT but you should point to some external SDKs folder to the generated SDK current folder El mar., 10 mar. 2020 a las 20:28, Brian Raymes (<[email protected]<mailto:[email protected]>>) escribió: Hi all, While testing Royale for a migration project, I started with building the latest SDK (update daily) with Ant. Linking the SDK to VSCode was all fine and well until I reported an issue was it was suggested I move to Maven. I now have a Maven project using the latest remote 0.9.7-SNAPSHOTs, but I still need a local SDK for code-completion, etc. The SDK produced with Ant was great. Code completion and error highlighting all work as expected. After switching to Maven, and building and linking the Distribution SDK, my code is now riddled with “errors” that are not really errors and code-completion only partially works. Here are a few issues I see with the Maven built SDK: 1. dispatchEvent(…any event…) is seen as an error because the Event is not an Object. · Implicit coercion of a value of type org.apache.royale.events.Event to an unrelated type Object.(1067) 1. addEventListener() / removeEventListener() are seen as errors in many cases. · Attempted access of inaccessible method addEventListener through a reference with static type Timer. 1. Code-completion doesn’t work for Jewel components, but does for properties. · <j:Bu will not suggest or auto-complete to <j:Button> · <j:Button la will auto complete label=”” Once I switch back to the Ant built SDK, all of these kinds errors go away and code-completion is working as expected. Thoughts? Brian -- Carlos Rovira http://about.me/carlosrovira
