Mailing-List: contact dev-help@flex.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@flex.apache.org
MIME-Version: 1.0
In-Reply-To: <D3EF67CF.74617%aharui@adobe.com>
References: <CANDXkxD6ceZ=VzUuZQu6+tB3bx2TFzO2xZ-u-JGXAtzHzq0yiA@mail.gmail.com>
 <D3E67526.73F7A%aharui@adobe.com> <CANDXkxBNh9BTz94nGsK=dgVMoQ8DRDw43zijXtc-gf1qSK7aYg@mail.gmail.com>
 <D3E90D41.74126%aharui@adobe.com> <CANDXkxD67qBhqGusRutwtsMbVMtZYxoxC7tmOD4bCpx-Nt4-BQ@mail.gmail.com>
 <D3EA84A6.7424C%aharui@adobe.com> <CANDXkxAKbQ3dXs8bY3kWFqauCd=7qT1oV7Qmr32ZMkvaZgF_nw@mail.gmail.com>
 <CANDXkxAve77P7Nx76QUeQSUGsQX=asEu=uaBPGSpC0HV6meS7g@mail.gmail.com>
 <D3EBCE0D.7432A%aharui@adobe.com> <CANDXkxBgvKxp4_ZZXRQO0zmucqGoHScqu0fK5xZUWOGFerNHxQ@mail.gmail.com>
 <D3EC4A89.74374%aharui@adobe.com> <CANDXkxC1p9RY-TqyY3vRJZ3aBOScekHYzYMDWq+iHW2c1cWwmQ@mail.gmail.com>
 <CANDXkxAiFAEqsK_DQ5bWqw+Un6rX2OZbaW2TuG8YWj1cJfmryQ@mail.gmail.com>
 <D3ECFA94.743F6%aharui@adobe.com> <D3ECFF60.74403%aharui@adobe.com>
 <CANDXkxD+yzpSPeFZ9n5XShGkJqjp-v1NSe68w6ssryDNG90nwA@mail.gmail.com>
 <D3ED0A97.74413%aharui@adobe.com> <CANDXkxA_P80gSFEoByijocXhsLCNoEzPf=VmVC1VEjSv8n6b7g@mail.gmail.com>
 <D3ED1582.74441%aharui@adobe.com> <CANDXkxDD4GZ0Z7=gHma1BH_yfNjw5pMPsOh6fA5dRFif4-Ri-w@mail.gmail.com>
 <CANDXkxB7djFvFZjtUzZ4Ob4LpLgAE5HmMm3YgWCGOWFWoSm1Cg@mail.gmail.com>
 <D3ED9411.7446E%aharui@adobe.com> <D3EE2F77.744F3%aharui@adobe.com>
 <CANDXkxDugctQs7_KfJunEKR04A0MJKYFuxFYo4vSenZCDHgUMg@mail.gmail.com>
 <CANDXkxA2kcPuxgrp-YixbjVdM3M8sEOp_hpaNNhpFYmxxayzDw@mail.gmail.com>
 <D3EEDF7D.74527%aharui@adobe.com> <CANDXkxBkmtOLqb7eVUnwcUtw3DqtvYs9U8nO=pi7_u-tPanYEg@mail.gmail.com>
 <CANDXkxDcu0deBw6XyH6MLFJ_hLi5+Wg2DcjzY9H=BDhjV+ovjg@mail.gmail.com>
 <D3EF545A.745ED%aharui@adobe.com> <CANDXkxCSXnsHXRku_bLQ2e+uJpiApD5ViCFzLqNDsXKkxynBKA@mail.gmail.com>
 <D3EF5E3C.74608%aharui@adobe.com> <CANDXkxBMTCyCPPFMiDvFJVrGnVBhexg81=jjSZ+=8+pbhqpY8g@mail.gmail.com>
 <D3EF67CF.74617%aharui@adobe.com>
From: Greg Dove <greg.dove@gmail.com>
Date: Sat, 3 Sep 2016 13:41:57 +1200
Message-ID: <CANDXkxCScHE+AAywPx3CuP2icRJvpKHUX7iU2BOQgoxNp8c0bA@mail.gmail.com>
Subject: Re: [FlexJX][Falcon] Binding support fixes/improvements
To: dev@flex.apache.org
Content-Type: multipart/alternative; boundary=001a11490a828d1c76053b908d3f
archived-at: Sat, 03 Sep 2016 01:42:06 -0000

--001a11490a828d1c76053b908d3f
Content-Type: text/plain; charset=UTF-8

Sound great. Yes, switching it back on at any point should remain a quick
fix option because it would simply make the other code act as a backstop.
I will take a quick look on Monday to see if I can find any problems in
small as-only projects where maybe the order of class definitions in the
swf could be important with the current approach.  Sounds a bit like a
theoretical edge case but you made me think about it.
Have a nice weekend.

-Greg
[sent from my phone]

On 3/09/2016 12:41 PM, "Alex Harui" <aharui@adobe.com> wrote:

>
>
> On 9/2/16, 5:10 PM, "Greg Dove" <greg.dove@gmail.com> wrote:
>
> >Yes, that was one of the major changes. I moved it alongside the other
> >implementation inside ClassDirectiveProcessor and was able to get more
> >consistent results.
> >I initially commented it out so I could get everything working using the
> >implements IEventDispatcher approach - which is the more general approach
> >that would work for all cases- in javascript, then added it back for both
> >targets as a 2nd output variation in a later commit, and it went to
> >ClassDirectorProcessor for swf.
> >
> >I was seeing 2 issues from the ASCompilationUnit implementation:
> >1. It was not always applying where it should have been (so in cases where
> >it could have been 'extends EventDispatcher' it would sometimes fall
> >through to the implements IEventDispatcher approach in
> >ClassDIrectiveProcessor anyway). I suspect this was threading-related, but
> >I am not so familiar with threads.
> >
> >2. Less of an issue, it could also apply an EventDispatcher base class in
> >static-only bindable cases, which is unnecessary, this I expect could have
> >been more easily fixed. I tried checking with hasModifier in the original
> >code, but I think that might not be available yet, iirc that caused an
> >error.
> >
> >So I moved everything to the ClassDirectiveProcessor, which had the other
> >implementation of bindable support in any case. Sorry, I thought I had
> >been
> >clear about that.
>
> Well, you probably did explain it, but it probably didn't stick in my head
> until we hit this last issue and I went looking for the code where I had a
> vague memory of hacking a base class.  I still don't know the compiler
> code so well that it is clear in my head what work should be done where.
> It looks like the original attempt to deal with [Bindable] for SWF was
> done in ClassDirectiveProcessor, but when I wanted to fix a bug in
> FalconJX I decided to fix it by hacking the AST since that drives
> everything in FalconJX.  The right answer may have been for me to move the
> code from ClassDirectiveProcessor into ASCompilationUnit and use the
> existing tests for needsEventDispatcher, then maybe we wouldn't have had
> the two issues above.
>
> Anyway, I think we have recorded the possibility that AST hacking might be
> a solution if we hit other problems.  I'm ok with leaving the code as is
> until we hit another problem but feel free to keep digging if you want.
>
> Thanks,
> -Alex
>
>

--001a11490a828d1c76053b908d3f--