flex-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nigel Magnay <nigel.mag...@gmail.com>
Subject Re: Weird bug when an unrelated skin class is present in an RSL (?!?)
Date Fri, 05 Sep 2014 19:26:20 GMT
I tell a lie, it's merged in both.

This might be a regression - it seems to work correctly if I use 4.10 SDK
in both projects, but break with 4.13 / 4.12

I'll try some more combinations tomorrow.



On Fri, Sep 5, 2014 at 8:22 PM, Nigel Magnay <nigel.magnay@gmail.com> wrote:

> It's using the defaults from FP, which are (I think) merged in the SWF and
> external in the RSL. I'll have to check to be sure though.
>
> What's weird (to me) though is the fact the class that breaks it in the
> RSL isn't ever used in the SWF (you can remove the reference to the library
> and it works fine). There's just something about even mentioning a class
> that seems to break something. :-?
>
>
> On Fri, Sep 5, 2014 at 8:11 PM, Alex Harui <aharui@adobe.com> wrote:
>
>> There isn't any special treatment of AS classes in the player.
>>
>> I didn't realize the class that makes a difference comes from the
>> experimental swc.  Is the experimental swc being loaded as an RSL or is it
>> merged in?
>>
>> -Alex
>>
>> On 9/5/14 11:55 AM, "Nigel Magnay" <nigel.magnay@gmail.com> wrote:
>>
>> >I always do... It's basically consistent; It seems that if the RSL
>> project
>> >uses a class that's in the 'experimental' SWC, then MenuBar totally
>> breaks
>> >in the main swf (and it doesn't seem to matter if that class is even used
>> >anywhere).
>> >
>> >Is there some special treatment by the flash runtime of classes declared
>> >to
>> >be itemrenderers or skins?
>> >
>> >
>> >
>> >
>> >On Fri, Sep 5, 2014 at 7:50 PM, Alex Harui <aharui@adobe.com> wrote:
>> >
>> >> Have you tried cleaning the RSL project so the RSL gets fully rebuilt?
>> >>
>> >> On 9/5/14 11:40 AM, "Nigel Magnay" <nigel.magnay@gmail.com> wrote:
>> >>
>> >> >It should be identical - it's launched from within flash builder, so
>> my
>> >> >understanding is it's just copied into the target directory...
>> >> >
>> >> >
>> >> >
>> >> >On Fri, Sep 5, 2014 at 7:16 PM, Alex Harui <aharui@adobe.com>
wrote:
>> >> >
>> >> >> Hi Nigel,
>> >> >>
>> >> >> I'm not asking about application domains, but the URL of the RSL
vs
>> >>the
>> >> >> URL of the main app.  If the actual string of the domain portion
of
>> >>the
>> >> >> URL is different in any way, the RSL could get loaded into a
>> sandbox.
>> >> >>
>> >> >> -Alex
>> >> >>
>> >> >> On 9/5/14 11:00 AM, "Nigel Magnay" <nigel.magnay@gmail.com>
wrote:
>> >> >>
>> >> >> >The RSL is loaded into the default app domain (it's a single
>> >>library in
>> >> >> >flash builder).
>> >> >> >
>> >> >> >It's odd.
>> >> >> >
>> >> >> >There's a totally minimal (I.E: it's got just 2 files) example
>> here:
>> >> >> >https://github.com/magnayn/flexbug
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> >On Fri, Sep 5, 2014 at 6:16 PM, Alex Harui <aharui@adobe.com>
>> wrote:
>> >> >> >
>> >> >> >> What is the load order of the RSLs?  You might need to
look at
>> >> >>generated
>> >> >> >> code to find the order.
>> >> >> >>
>> >> >> >> And just to double-check, "same domain" means full domain:
>> >> >> >>"rsls.foo.com"
>> >> >> >> is a different domain than "www.foo.com" and even the
IP address
>> >>or
>> >> >> >> "localhost" is different than other domain names mapped
to the
>> >>same
>> >> >> >>place.
>> >> >> >>
>> >> >> >> Are you running a debug build with debug code in the RSLs?
 No
>> >>other
>> >> >> >> interesting output in the console?
>> >> >> >>
>> >> >> >> -Alex
>> >> >> >>
>> >> >> >> On 9/5/14 9:36 AM, "Nigel Magnay" <nigel.magnay@gmail.com>
>> wrote:
>> >> >> >>
>> >> >> >> >I am. It does feel like a class isolation issue though.
>> >> >> >> >
>> >> >> >> >
>> >> >> >> >On Fri, Sep 5, 2014 at 5:32 PM, Alex Harui <aharui@adobe.com>
>> >> wrote:
>> >> >> >> >
>> >> >> >> >> Are you loading the RSL from the same domain
as the app?
>> >> >> >> >>
>> >> >> >> >> -Alex
>> >> >> >> >>
>> >> >> >> >> On 9/5/14 4:47 AM, "Nigel Magnay" <nigel.magnay@gmail.com>
>> >>wrote:
>> >> >> >> >>
>> >> >> >> >> >Even more strangely, I've narrowed it down
still more. The
>> >>mere
>> >> >> >> >>inclusion
>> >> >> >> >> >of this ItemRenderer is sufficient to break
the menubar:
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >>
>> >> >>
>> >>
>> >>
>> https://github.com/magnayn/flexbug/blob/master/buglib/src/buglib/componen
>> >> >> >> >>t
>> >> >> >> >> >s/skins/MenuCoreItemRenderer2.as
>> >> >> >> >> >
>> >> >> >> >> >If you comment out the use of MenuEvent,
it works.
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >> >On Fri, Sep 5, 2014 at 12:14 PM, Nigel Magnay
>> >> >> >><nigel.magnay@gmail.com>
>> >> >> >> >> >wrote:
>> >> >> >> >> >
>> >> >> >> >> >> So, I've narrowed this down to a test
project - see
>> >> >> >> >> >> https://github.com/magnayn/flexbug
>> >> >> >> >> >>
>> >> >> >> >> >> I have a project that uses the "Spark
Menubar". It's
>> >>basically
>> >> >>the
>> >> >> >> >> >>example
>> >> >> >> >> >> given in the documentation:
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >>
>> >> https://github.com/magnayn/flexbug/blob/master/buglet/src/buglet.mxml
>> >> >> >> >> >>
>> >> >> >> >> >> This all works fine, in isolation. However.
If the project
>> >> >> >>includes
>> >> >> >> >> >> 'buglib' as an RSL, the menu stops working.
>> >> >> >> >> >>
>> >> >> >> >> >> buglib contains only 1 file, which is
a skin:
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >>
>> >> >>
>> >>
>> >>
>> https://github.com/magnayn/flexbug/blob/master/buglib/src/buglib/componen
>> >> >> >> >> >>ts/skins/TestSkin.mxml
>> >> >> >> >> >>
>> >> >> >> >> >> The skin does not mark in metadata the
HostComponent (which
>> >> >>would
>> >> >> >>be
>> >> >> >> >>a
>> >> >> >> >> >> spark list). The skin is not referenced
*anywhere *from the
>> >> >>MXML.
>> >> >> >> >> >>
>> >> >> >> >> >> If the library is included instead as
'merged into code',
>> >> >> >>everything
>> >> >> >> >>is
>> >> >> >> >> >> fine (even if the skin is referenced,
to prevent it being
>> >> >> >>stripped).
>> >> >> >> >> >>
>> >> >> >> >> >> And, here's where it gets even weirder
:
>> >> >> >> >> >> If you remove the statement
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >>itemRenderer="spark.components.itemRenderers.MenuBarItemRenderer"
>> >> >> >> >> >>
>> >> >> >> >> >> From the TestSkin.mxml (*even though
it is not being used
>> >> >> >>anywhere in
>> >> >> >> >> >>the
>> >> >> >> >> >> project*), it all works fine again!
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >> Ideas? Have I just missed something?
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >>
>> >> >>
>> >>
>> >>
>>
>>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message