Thanks Chris, that's helpful. Do you have any thoughts on how to manage two different sets of libraries, one for SWF compile, one for JS compile?
Thanks, -Alex On 12/5/16, 12:44 AM, "Christofer Dutz" <christofer.d...@c-ware.de> wrote: >Hi Alex, > >sorry for not responding on that last email ☹... must have missed that >one ☹ > >For the general functionality of the flex-tool-groups, I simply use the >basic Java service registry functionality. >This allows to lookup a list of instances with a symbolic name. You can >add more implementations to the list, by providing a jar with a simple >text file named: “META-INF/services/org.apache.flex.tools.FlexToolGroup”. >This file only contains class names. Each row will be treated as an >alternate implementation. >So if I want to get all “tool-groups”, all I have to do, is to ask the >Service registry to give me all “org.apache.flex.tools.FlexToolGroup”. >In my case I made sure I defined a universal interface that all >implementations would implement, so I know that I can simply cast each >instance I get back to that base interface. >In my case in order to add a new compiler called “foo”, all I would do, >is to add a jar with a class implementing FlexToolGroup and to make >“getName” return “foo”. > >Currently I would have to extend the Enum JSOutputType to support “foo”, >which doesn’t make it that extensible. > >Right now I have a template for each type of compilation and a matching >“Mojo”. All compiler options go into the configuration file for that >particular Mojo. Additionally, the Mojo is able to customize the default >values used to generate the template. For example, the getDefines method >returns different values for CompileASMojo and for CompileJSMojo. > >All I have to do then, is to provide the template path to the compiler >using the default “execute” method of any of the “FlexTool” >implementations and simply provide an array of Strings which equals the >normal array of Strings the compiler gets in the Ant world. > >Hope that explains things a little more. > >Chris > > >Am 02.12.16, 19:13 schrieb "Alex Harui" <aha...@adobe.com>: > > I'm about to start working on this. I haven't actually tried anything > yet, but my thinking is that there are some interesting issues >regarding > configuration. Typically, when compiling an application, you define a > bunch of compiler options, but for a compilation that should produce >two > different outputs, you might need two different sets of compiler >options. > How should that be managed? > > For example, you might want a different library-path for the SWF >compile > vs the JS compile. Or set COMPILE::SWF=true for the SWF build and > COMPILE::SWF=false for the JS compile. > > One possibility is that all platform-specific options have to be >specified > in -config.xml files and not on the command line. Then for any > -load-config option, the JS compile will look for a js-prefixed file. > So > instead of using compile-as-config.xml and compile-js-config.xml, we >would > use compile-config.xml and js-compile-config.mxml and you would >specify > -load-config=compile-config.xml as the compiler option. > > Another possibility is that the SWF compile would always set > COMPILE::SWF=true and COMPILE::JS=false and the JS compile would do >the > opposite. The JS compile would see if -js-library-path was specified >and > if not, use the -library-path. The weird thing about this approach is > that the SWF compiler needs to support a -js-library-path option >otherwise > the SWF compiler's configuration checker will throw an error. Adding > options that the SWF compiler must ignore doesn't seem right to me. > > Any other ideas? Thoughts? > -Alex > > > On 10/7/16, 8:09 AM, "Alex Harui" <aha...@adobe.com> wrote: > > >I don't know enough about how Flex Tool Groups work to form an >opinion. > >Can you provide more details and explain how folks would configure >things > >on the command-line? > > > >AIUI, -js-output-type is sort of overloaded right now, because the > >FLEXJS_DUAL dictates the output of a SWF. In my plan to combine >things, I > >would introduce a new configuration parameter like -output-type or > >-targets or something like that. I supposed it could be a list of > >compiler jars to run. But -js-output-type doesn't currently go >through > >different jars for JSC, NODE, and FLEXJS. That parameter just sets >up a > >few configuration changes within the output classes. > > > >I'm pretty sure Schmalle imagined a true plug-in architecture for the > >emitters as well. Not sure we want to take all of that on right >now, but > >we might want to consider that as well. > > > >-Alex > > > >On 10/7/16, 12:14 AM, "Christofer Dutz" <christofer.d...@c-ware.de> >wrote: > > > >>I would like to propose a different approach: > >> > >> > >>Unless these approaches led to a dead end and there is no solution >down > >>that path, I would rather suggest to refactor the compiler to allow > >>dynamic adding of alternate compilers. Currently the supported >compilers > >>are limited by the JsOutputType enum. If we however used something >like > >>the Java service thing I used for the Flex Tool Groups, we could >allow > >>new experiments to start in dedicated Jars and we wouldn't have >such a > >>mixture in case someone wants to try something out. > >> > >> > >>Chris > >> > >>________________________________ > >>Von: Alex Harui <aha...@adobe.com> > >>Gesendet: Freitag, 7. Oktober 2016 00:25:30 > >>An: dev@flex.apache.org > >>Betreff: Re: [FALCONJX] Combining SWF and JS compilers (was Re: AW: > >>[FalconJX][FlexJS] COMPJSC and Build order) > >> > >>+1. Bonus points for making a tag and/or branch before deleting. > >> > >>-Alex > >> > >>On 10/6/16, 3:02 PM, "Josh Tynjala" <joshtynj...@gmail.com> wrote: > >> > >>>Makes sense to me. I say we simply delete it. No need to transfer >to > >>>another folder. It'll still be in the repository's history. > >>> > >>>- Josh > >>> > >>>On Thu, Oct 6, 2016 at 2:55 PM, Carlos Rovira > >>><carlos.rov...@codeoscopic.com > >>>> wrote: > >>> > >>>> I want to propose the same. I something was an experiment and is >no > >>>>more > >>>> developed. I think it should go to some folder that make people >avoid > >>>> confusion about what code is valid and what is left behind. > >>>> > >>>> 2016-10-06 23:12 GMT+02:00 Christofer Dutz > >>>><christofer.d...@c-ware.de>: > >>>> > >>>> > I stumble over tons of VF2JS classes and think it would be >better to > >>>>move > >>>> > stuf like that to some sort of attic. What do you think? > >>>> > > >>>> > > >>>> > Chris > >>>> > > >>>> > ________________________________ > >>>> > Von: Josh Tynjala <joshtynj...@gmail.com> > >>>> > Gesendet: Donnerstag, 6. Oktober 2016 17:13:49 > >>>> > An: dev@flex.apache.org > >>>> > Betreff: Re: [FALCONJX] Combining SWF and JS compilers (was >Re: AW: > >>>> > [FalconJX][FlexJS] COMPJSC and Build order) > >>>> > > >>>> > JSC is meant to be purely an ActionScript to JavaScript >transpiler > >>>> without > >>>> > any frameworks. By default, it doesn't export an HTML file, >but it > >>>>will > >>>> > optionally support custom HTML templates in 0.8.0. It is >exposed > >>>>through > >>>> > the js/bin/asjsc executable, where it loads the > >>>>frameworks/js-config.xml > >>>> > configuration. js-config.xml references js.swc to give >ActionScript > >>>> access > >>>> > to browser APIs. > >>>> > > >>>> > NODE generates an index.js that bootstraps things for Node.js. >It is > >>>> > exposed through the js/bin/asnodec executable, which it loads >the > >>>> > frameworks/node-config.xml configuration. In addition to >js.swc, > >>>> > node-config.xml references node.swc to give ActionScript >access to > >>>> Node.js > >>>> > APIs. > >>>> > > >>>> > As far as I know, AMD and VF2JS are no longer maintained. I >assume > >>>>AMD > >>>> > tried to output AMD modules instead of goog modules. I >remember Alex > >>>>or > >>>> > someone mentioning that VF2JS had something to do with the >original > >>>>Flex > >>>> > framework, but I don't know the details. > >>>> > > >>>> > - Josh > >>>> > > >>>> > On Thu, Oct 6, 2016 at 12:10 AM, Christofer Dutz < > >>>> > christofer.d...@c-ware.de> > >>>> > wrote: > >>>> > > >>>> > > Hi Alex, > >>>> > > > >>>> > > > >>>> > > yesterday I stumbled over this flexjs-dual output type while > >>>>looking > >>>> for > >>>> > > the correct settings to buid a pure JS app. Would it be >possible > >>>>for > >>>> you > >>>> > > guys to give a short summary of what the different output >types > >>>> actually > >>>> > > are? The enum doesn't contain any documentation on this and I > >>>>guess > >>>> this > >>>> > > would be really helpful. > >>>> > > > >>>> > > > >>>> > > So far I see these output types: > >>>> > > > >>>> > > AMD > >>>> > > FLEXJS > >>>> > > GOOG > >>>> > > VF2JS > >>>> > > FLEXJS_DUAL > >>>> > > JSC > >>>> > > NODE > >>>> > > > >>>> > > And I guess I only used no value and FLEXJS > >>>> > > > >>>> > > > >>>> > > Chris > >>>> > > > >>>> > > ________________________________ > >>>> > > Von: Alex Harui <aha...@adobe.com> > >>>> > > Gesendet: Donnerstag, 6. Oktober 2016 07:45:48 > >>>> > > An: dev@flex.apache.org > >>>> > > Betreff: [FALCONJX] Combining SWF and JS compilers (was Re: >AW: > >>>> > > [FalconJX][FlexJS] COMPJSC and Build order) > >>>> > > > >>>> > > Fred Thomas did some work in this area about a year ago. In >the > >>>> > > flex-oem-compiler module that FB (and maybe other IDEs) use >to > >>>>talk > >>>>to > >>>> > the > >>>> > > compiler, he added a FLEXJS_DUAL -js-output-type. Not sure >how > >>>>well it > >>>> > > works. > >>>> > > > >>>> > > Thinking about this some more we'd have to have the same > >>>>configuration > >>>> > > options available to both compilers which might be a bit >strange. > >>>>Or > >>>> > > maybe we can convince the compilers to not complain about >unknown > >>>> config > >>>> > > parameters. > >>>> > > > >>>> > > We'd have to decide on how to reset the library-path for each > >>>>compile. > >>>> > > The JS compile might use different SWCs than the SWF compile. > >>>> > > > >>>> > > We'd have to select a few conditional compile options that >would > >>>>be > >>>> > > different for each compiler. For example, COMPILE::SWF >would be > >>>>true > >>>> for > >>>> > > SWF compiling and false for JS compiling and vice versa, and >maybe > >>>> > finding > >>>> > > those params on the command-line would have no effect since >they > >>>>would > >>>> be > >>>> > > dictated by the compiler. > >>>> > > > >>>> > > Thoughts? > >>>> > > -Alex > >>>> > > > >>>> > > On 10/2/16, 1:45 PM, "Josh Tynjala" <joshtynj...@gmail.com> >wrote: > >>>> > > > >>>> > > >That would be ideal! > >>>> > > > > >>>> > > >- Josh > >>>> > > > > >>>> > > >On Oct 1, 2016 10:47 PM, "Alex Harui" <aha...@adobe.com> >wrote: > >>>> > > > > >>>> > > >> One more thought on this: now that COMPJSC can more or >less > >>>>build > >>>> its > >>>> > > >>own > >>>> > > >> output instead of relying on COMPC to package its pile of >.js > >>>>files, > >>>> > it > >>>> > > >> might be worth experimenting with combining Falcon and >FalconJX > >>>>so > >>>> > COMPC > >>>> > > >> can produce a SWC or a SWC with JS files based on some > >>>>configuration > >>>> > > >> parameter. Then there would only be one compiler that >produces > >>>>SWFs > >>>> > or > >>>> > > >>JS > >>>> > > >> based on some -output-type flag. > >>>> > > >> > >>>> > > >> Thoughts? > >>>> > > >> -Alex > >>>> > > >> > >>>> > > >> On 10/1/16, 10:18 PM, "Alex Harui" <aha...@adobe.com> >wrote: > >>>> > > >> > >>>> > > >> >Hi Chris, > >>>> > > >> > > >>>> > > >> >When I read this, I realized I already pushed the >changes when > >>>>I > >>>> > pushed > >>>> > > >> >some other changes yesterday. If the Maven build didn't >blow > >>>>up, > >>>> it > >>>> > is > >>>> > > >> >probably because it is using its own >compile-xx-config.xml > >>>>files so > >>>> > is > >>>> > > >> >still generating a pile of .js files and packaging them >up on > >>>>the > >>>> SWF > >>>> > > >> >COMPC run. > >>>> > > >> > > >>>> > > >> >-Alex > >>>> > > >> > > >>>> > > >> >On 10/1/16, 6:10 AM, "Christofer Dutz" > >>>><christofer.d...@c-ware.de> > >>>> > > >>wrote: > >>>> > > >> > > >>>> > > >> >>Hi Alex, > >>>> > > >> >> > >>>> > > >> >> > >>>> > > >> >>so I guess ideally this change should be done on a >feature > >>>>branch, > >>>> > so > >>>> > > >>I > >>>> > > >> >>can sort out the Maven issues and we'll merge that back >as > >>>>soon as > >>>> > > >>all is > >>>> > > >> >>working. I would like to ask you to create a > >>>> > > >>"feature-autobuild/"-branch > >>>> > > >> >>for that. Just give me a short note what branch the >stuff is > >>>>in > >>>> and > >>>> > > >>I'll > >>>> > > >> >>try to sort out the Maven issues. > >>>> > > >> >> > >>>> > > >> >> > >>>> > > >> >>Chris > >>>> > > >> > > >>>> > > >> > >>>> > > >> > >>>> > > > >>>> > > > >>>> > > >>>> > >>>> > >>>> > >>>> -- > >>>> > >>>> Carlos Rovira > >>>> Director General > >>>> M: +34 607 22 60 05 > >>>> http://www.codeoscopic.com > >>>> http://www.avant2.es > >>>> > >>>> > >>>> Este mensaje se dirige exclusivamente a su destinatario y puede > >>>>contener > >>>> información privilegiada o confidencial. Si ha recibido este >mensaje > >>>>por > >>>> error, le rogamos que nos lo comunique inmediatamente por esta >misma > >>>>vía y > >>>> proceda a su destrucción. > >>>> > >>>> De la vigente Ley Orgánica de Protección de Datos (15/1999), le > >>>>comunicamos > >>>> que sus datos forman parte de un fichero cuyo responsable es > >>>>CODEOSCOPIC > >>>> S.A. La finalidad de dicho tratamiento es facilitar la >prestación del > >>>> servicio o información solicitados, teniendo usted derecho de >acceso, > >>>> rectificación, cancelación y oposición de sus datos dirigiéndose >a > >>>>nuestras > >>>> oficinas c/ Paseo de la Habana 9-11, 28036, Madrid con la > >>>>documentación > >>>> necesaria. > >>>> > >> > > > > >
