Re: Mustella failures

[Erik de Bruin]

> Was finally able to get the mobile tests to run the mobile tests and can 
> repeat the two failures. I had to revert to 11.1 and AIR 3.1 - are we sure 
> that these test have been tested on laters versions of air?

AIR 3.7 is the oldest version of the AIR SDK on the Mustella VM. It
currently runs the tests on 3.7, 3.8 and 3.9.

Then again, I may have messed up the setup of the tests - wouldn't be
the first time. Feel free to check me, please.

EdB



-- 
Ix Multimedia Software

Jan Luykenstraat 27
3521 VB Utrecht

T. 06-51952295
I. www.ixsoftware.nl

CursorManager

[Harbs]

There's an old issue with Flex that CursorManager does not provide access to 
the current cursor to customize its look and behavior.
(Here's one old blog post on the topic: 
http://jessewarden.com/2009/01/making-a-cooler-cursor-in-flex.html)

I currently have a need to create a dynamic cursor, and I thought I'd make the 
cursor available in CursorManager so people don't need to monkey patch things 
every time they need this. Is there any reason this would be a bad idea? Was 
there a valid reason it was made private in the original design?

Harbs

Re: CursorManager

[Alex Harui]

I don't remember any reason otoh. I know cursor can change as mouse goes over 
text. Can you cache cursor without causing memory leak?




Sent via the PANTECH Discover, an AT&T 4G LTE smartphone.

Harbs  wrote:


There's an old issue with Flex that CursorManager does not provide access to 
the current cursor to customize its look and behavior.
(Here's one old blog post on the topic: 
http://jessewarden.com/2009/01/making-a-cooler-cursor-in-flex.html)

I currently have a need to create a dynamic cursor, and I thought I'd make the 
cursor available in CursorManager so people don't need to monkey patch things 
every time they need this. Is there any reason this would be a bad idea? Was 
there a valid reason it was made private in the original design?

Harbs

RE: Debug ActionScript workers

[Frédéric THOMAS]

Any Progress on this ?

Just tried to overlay the Air SDK compiler on top of the Apache Flex SDK to
get it working in IntelliJ but I  have only trace statements displayed, it
can't stop on breakpoints.

We're now at AIR 3.8 in the official release and workers can be debugged in
FB 4.7 since 3.4, it would be fair other IDEs be able to leverage this
feature.

-Fred

-Message d'origine-
De : Alex Harui [mailto:aha...@adobe.com] 
Envoyé : jeudi 7 mars 2013 18:14
À : dev@flex.apache.org
Objet : Re: Debug ActionScript workers




On 3/7/13 4:28 AM, "Alexander Doroshko" 
wrote:

> 
> Alex, did you get any any response?
> In my opinion fully-functional debugging tool is a must for Apache Flex
SDK.
I got the ok from the business folks.  Now I have to actually find the code
and get it cleared legally.  It isn't high on my priority list at this
point, but I think there's a good chance it will make it into Apache Flex
eventually.

--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui

Re: Build failed in Jenkins: flex-sdk_mustella-mobile #128

[OmPrakash Muppirala]

I checked in the fixes the skins for the CalloutButton last night.  But
this failure seems to be related, but in a different area.  Should I take a
look at it or is someone else looking at this?

Thanks,
Om


On Sun, Sep 8, 2013 at 8:30 AM,  wrote:

> See 
>
> --
> [...truncated 16041 lines...]
>  [echo] have_air_adt_jar is true
>  [echo] have_air_android_runtime is ${have_air_android_runtime}
>  [echo] need_air_android is ${need_air_android}
>  [echo] need_air_ios is ${need_air_ios}
>  [echo] need_to_show_air_ios_fail_message is
> ${need_to_show_air_ios_fail_message}
>  [echo] need_to_fetch_air_ios is ${need_to_fetch_air_ios}
>
> show_air_ios_fail_message:
>
> show_air_android_fail_message:
>
> fetch_air_ios:
>
> fetch_qnx_sdk:
>
> echo-browser:
>
> echo-apollo:
>  [echo] use_apollo=true
>  [echo] apollo_location is C:\ApacheFlex\dependencies\AdobeAIRSDK\3.7
>
> setup:
>  [echo] player is
> C:\ApacheFlex\dependencies\FlashPlayer_Debug\flashplayer11_1r102_55_win_sa_debug_32bit.exe
>  [echo] fileset:
> mobile\SkinnablePopupContainerSK\integration\SPUC_KeyboardEffects_CalloutButton.mxml;mobile\SkinnablePopupContainerSK\integration\SPUC_KeyboardEffects_CalloutButton_List.mxml
>
> setbuildID:
>  [echo] Target file was:
> C:/jenkins_slave/workspace/flex-sdk_mustella-mobile/mustella/successfulBuild.properties
>  [echo] ${server}
>
> setHostName:
>
> getConfigId:
>
> getActualRunId:
>
> getRunId:
>
> compilemustellaswc:
>  [exec] Loading configuration file
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\flex-config.xml
>  [exec]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\mustella\mustella.swc
> (90910 bytes)
>
> realCompile:
>
> build_shell_set:
>  [echo] shell_file_mxml_equivs is
> ,mobile/SkinnablePopupContainerSK/integration/SPUC_KeyboardEffects_CalloutButton.sh,mobile/SkinnablePopupContainerSK/integration/SPUC_KeyboardEffects_CalloutButton_List.sh
>  [echo] mxml_equiv_shells is ${mxml_equiv_shells}
>  [echo] tmp.sdk.mustella.scripts2 is ;
>
> chmod_shells:
>  [echo] changing user shell files to executable
>
> shells:
>  [exec] done with pre compile step
>  [java] exclude_filename:
> C:/jenkins_slave/workspace/flex-sdk_mustella-mobile/mustella/tests/ExcludeList${os}.txt
>  [java] os_version: ${os_version}
>  [java] target_os_name: android
>  [java] device_name: win
>  [java] result_include: -includes=SendResultsToRunner
>  [java] created
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\mustella\as3\src\mustella\MobileConfig.as
>  [java] nothing left to do
>  [java] Choosing local runner bitmap save
>  [java] okey doke, going to compile
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\mustella\tests\mobile\SkinnablePopupContainerSK\SWFs\SkinnablePopUpContainerSKEffects.mxml
>  [java] Loading configuration file
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\airmobile-config.xml
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(38):
> Warning: Cannot resolve namespace qualified type 'BorderDataNavigator' in
> CSS selector 'BorderDataNavigator'
>  [java]
>  [java]
>  [java]
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(29):
> Warning: The style 'cornerRadius' is only supported by type
> 'spark.components.ButtonBarButton' with the theme(s) 'spark'.
>  [java]
>  [java]
>  [java]
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(69):
> Warning: The style 'cornerRadius' is only supported by type
> 'spark.components.ButtonBarButton' with the theme(s) 'spark'.
>  [java]
>  [java]
>  [java]
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(38):
> Warning: Cannot resolve namespace qualified type 'BorderDataNavigator' in
> CSS selector 'BorderDataNavigator'
>  [java]
>  [java]
>  [java]
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(38):
> Warning: Cannot resolve namespace qualified type 'BorderDataNavigator' in
> CSS selector 'BorderDataNavigator'
>  [java]
>  [java]
>  [java]
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(38):
> Warning: Cannot resolve namespace qualified type 'BorderDataNavigator' in
> CSS selector 'BorderDataNavigator'
>  [java]
>  [java]
>  [java]
>  [java]
> C:\jenkins_slave\workspace\flex-sdk_mustella-mobile\frameworks\libs\experimental.swc$defaults.css(38):
> Warning: Cannot resolve namespace qualified type 'BorderDataNavigator' in
> CSS selector 'BorderDataNavigator'
>

Re: Build failed in Jenkins: flex-sdk_mustella-mobile #128

[Justin Mclean]

Hi,

> I checked in the fixes the skins for the CalloutButton last night.  But this 
> failure seems to be related, but in a different area.  Should I take a look 
> at it or is someone else looking at this?
Not seeing this issue in the latest build with the fix I checked in this 
morning.

Thanks,
Justin

Re: Mustella failures

[Justin Mclean]

Hi,

>   if ("softKeyboardRect" in FlexGlobals.topLevelApplication)
> 
> isn't right because the application's softkeyboard rect is an mx_internal
> variable and "in" only tests public variables. 

Thanks for that, fixed it by  changing it to be a try catch, the tests now pass.

> Mustella was correct again, but you are right that it is hard to debug
> these tests.  Did you ever get the debugger to work on a mustella test?
Not yet no - that would certainly make it easier. 

What's the reason that the test swfs can't be run in FB (or other IDE)?

Thanks,
Justin

Re: Mustella failures

[Alex Harui]

On 9/8/13 5:16 PM, "Justin Mclean"  wrote:

>Hi,
>
>>  if ("softKeyboardRect" in FlexGlobals.topLevelApplication)
>> 
>> isn't right because the application's softkeyboard rect is an
>>mx_internal
>> variable and "in" only tests public variables.
>
>Thanks for that, fixed it by  changing it to be a try catch, the tests
>now pass.
>
>> Mustella was correct again, but you are right that it is hard to debug
>> these tests.  Did you ever get the debugger to work on a mustella test?
>Not yet no - that would certainly make it easier.
OK, let's try to get you set up.  Let us know what errors or failure
conditions you are hitting.
> 
>
>What's the reason that the test swfs can't be run in FB (or other IDE)?
AFAIK, they can.  I don't do it because there are so many test SWFs it is
way more efficient to use FDB (for me).  Supposedly there is a way to
create a generic project in FB, launch it in the debugger, but somehow it
doesn't run and you launch the mustella SWF instead.  Not sure of the
details, but I think other folks on the list know how to do this.  I can't
imagine setting up an FB project for every test SWF.

-Alex

Re: Debug ActionScript workers

[Alex Harui]

I have the code, but it has to be legally cleared and I haven't made it a
priority.  I'm doing IP clearance on BlazeDS right now and it might also
require that some other files get donated and I'll bundle this with it.

-Alex

On 9/8/13 7:00 AM, "Frédéric THOMAS"  wrote:

>Any Progress on this ?
>
>Just tried to overlay the Air SDK compiler on top of the Apache Flex SDK
>to
>get it working in IntelliJ but I  have only trace statements displayed, it
>can't stop on breakpoints.
>
>We're now at AIR 3.8 in the official release and workers can be debugged
>in
>FB 4.7 since 3.4, it would be fair other IDEs be able to leverage this
>feature.
>
>-Fred
>
>-Message d'origine-
>De : Alex Harui [mailto:aha...@adobe.com]
>Envoyé : jeudi 7 mars 2013 18:14
>À : dev@flex.apache.org
>Objet : Re: Debug ActionScript workers
>
>
>
>
>On 3/7/13 4:28 AM, "Alexander Doroshko" 
>wrote:
>
>> 
>> Alex, did you get any any response?
>> In my opinion fully-functional debugging tool is a must for Apache Flex
>SDK.
>I got the ok from the business folks.  Now I have to actually find the
>code
>and get it cleared legally.  It isn't high on my priority list at this
>point, but I think there's a good chance it will make it into Apache Flex
>eventually.
>
>--
>Alex Harui
>Flex SDK Team
>Adobe Systems, Inc.
>http://blogs.adobe.com/aharui
>

Re: Mustella failures

[Justin Mclean]

HI,

> OK, let's try to get you set up.  Let us know what errors or failure
> conditions you are hitting.
No error occurs, swf launches but is blank and debugger hangs. Not really much 
to go on.

> I can't imagine setting up an FB project for every test SWF.
No but you could set up as required for tests that fail. Most test pass most of 
the time.

Thanks,
Justin

Re: Mustella failures

[Alex Harui]

On 9/8/13 8:49 PM, "Justin Mclean"  wrote:

>HI,
>
>> OK, let's try to get you set up.  Let us know what errors or failure
>> conditions you are hitting.
>No error occurs, swf launches but is blank and debugger hangs. Not really
>much to go on.
Puzzling.  How are you launching?  From Finder or FB or command-line?
Does the Standalone Debugger launch or some browser?

-Alex

Re: Mustella failures

[Justin Mclean]

Hi,

> From Finder or FB or command-line?
command line.

$fdb
(fdb) run SkinnablePopUpContainerSKEffects.swf
Attempting to launch and connect to Player using URL
SkinnablePopUpContainerSKEffects.swf

This result in the same blank swf:
fdb ./tests/mobile/SkinnablePopupContainerSK/ss/*.swf

Attempting to launch and connect to Player using URL
./tests/mobile/SkinnablePopupContainerSK/swfs/SkinnablePopUpContainerSKEffects.swf

Also running fdb just trying run and running the swf almost works
(fdb) run
Waiting for Player to connect
Player connected; session starting.
Set breakpoints and then type 'continue' to resume the session.
[SWF] SkinnablePopUpContainerSKEffects.swf - 4,192,841 bytes after decompression

But the tests time out before you have a chance to set breakpoints.

> Does the Standalone Debugger launch or some browser?
Standalone debugger starts but is blank, with the correct path to the swf  is 
in the title.

Justin

Re: Mustella failures

[Alex Harui]

Ah, sorry.  Forgot that this is a mobile test so you can't just launch it
in FDB because it needs AIR.

I have tried a couple of different workflows for mobile.  For this test, I
started fdb and just typed run.  Then I went to the mustella folder, typed
./mini_run.sh -rerun -timeout-6 and whatever you used to run the test.
 Mustella will skip the compile and run the test and now you have 60
seconds before it quits the test to debug it.

Sometimes, 60 seconds isn't long enough, so I do the above and then in
Finder I go and find the -app.xml file in the SWFs folder and make a copy
of it in that 60 second window.  It may just be named .xml instead of
-app.xml.  Then I can go run fdb again, type run, then go to another
window and run adl with the correct params and then I should be able to
spend all day in the debugger.

I think the following command line should work on your mac:

$AIR_HOME/bin/adl -runtime $AIR_HOME/runtimes/air/mac
adl_extras=-screensize 640x960:640x960 -profile mobileDevice -XscreenDPI
240 

FWIW, tests in the AIR run can be run with a keep option which saves the
-app.xml, but for some reason, mobile tests can't.

Hope that works for you,
-Alex


On 9/8/13 9:09 PM, "Justin Mclean"  wrote:

>Hi,
>
>> From Finder or FB or command-line?
>command line.
>
>$fdb
>(fdb) run SkinnablePopUpContainerSKEffects.swf
>Attempting to launch and connect to Player using URL
>SkinnablePopUpContainerSKEffects.swf
>
>This result in the same blank swf:
>fdb ./tests/mobile/SkinnablePopupContainerSK/ss/*.swf
>
>Attempting to launch and connect to Player using URL
>./tests/mobile/SkinnablePopupContainerSK/swfs/SkinnablePopUpContainerSKEff
>ects.swf
>
>Also running fdb just trying run and running the swf almost works
>(fdb) run
>Waiting for Player to connect
>Player connected; session starting.
>Set breakpoints and then type 'continue' to resume the session.
>[SWF] SkinnablePopUpContainerSKEffects.swf - 4,192,841 bytes after
>decompression
>
>But the tests time out before you have a chance to set breakpoints.
>
>> Does the Standalone Debugger launch or some browser?
>Standalone debugger starts but is blank, with the correct path to the swf
> is in the title.
>
>Justin

Re: [FlexJS] General questions

[jude]

Yes, I agree. What might alleviate that pain would be:

• move all the styles metadata into the skin instead of the component. have
the component declare skin styles like we declare skin parts, so
[SkinStyle(name="color", required=true)]
• or allow styles to be declared in the skin and make the compiler add
those styles to the list of styles available to that component type for
code completion and compilation
• or add a compiler flag to ignore attributes it is not aware of (highlight
in a different color?) and treat those unknown attributes as styles.
• or add a styles proxy object on the component for styles not known to the
component


   


I would prefer the above to inline, "color:red;cornerRadius:8".

The more I think about it MXML should be a generic XML description of an
application. Since we want MXML to output to other languages things like
binding expressions and inline name value pairs like,
style="fontFamily:Arial;fontWeight:bold" are getting away from the one of
the reasons that XML was created; that being readability.

I think another feature of an XML spec is that it defines guidelines for
parsers. E4X is an advanced notation to get right to the information you
need in an XML document. With, style="color:yellow;fontFamily:Times New
Roman" you would need to use E4X to get the attribute and then another
expression engine, RegExp, to get the names and values you need.

I'm not saying no to inline styles but to keep in mind MXML's roots. In
time, as MXML is adopted, the tooling and the developer base will grow as
well and I would think we'd want a distributable document to be as portable
and parsable as possible (say that 3 times fast).

Are you using the style property as a proxy to the CSS class now?

MXML,



HTML,







On Tue, Sep 3, 2013 at 7:03 PM, Alex Harui  wrote:

>
>
> On 9/3/13 4:55 PM, "OmPrakash Muppirala"  wrote:
>
> >On Tue, Sep 3, 2013 at 4:38 PM, Alex Harui  wrote:
> >
> >> Good question.  I added a FalconJX and FlexJS component to JIRA.  Please
> >> submit patches there.  There is already a Falcon component.
> >>
> >> In theory, if you have problems compiling a SWF that FB/MXMLC compiled,
> >> then file the bug against Falcon.  If Falcon compiled the SWF but you
> >>get
> >> an error creating the JS version, file the bug against FalconJX, and if
> >> there is a bug in the AS or JS code, file the bug against FlexJS.
> >>
> >> The styling implementation on JS is not yet complete, so you can't use
> >> non-standard CSS styles in JS right now.  But the goal is to support the
> >> appropriate CSS styles in the SWF since those styles should "just work"
> >>in
> >> the browser.
> >>
> >> In-line styles (like ) isn't supported yet
> >> either.  I'm on the fence about whether to keep adding styles/properties
> >> or just use HTML's styles object (like ).
> >> Some properties like "width" also support being set in CSS.  Opinions
> >> welcome.
> >>
> >
> >It would be nice if we retained the current Flex way of inlining
> >styles/properties, please.
> I assume your main reason is compatibility?  That's definitely a strong
> argument, but here's the counter-argument:
>
> IMO, a pain point in Flex was that styles had to be specified on the
> component when the style implementation was actually in the skin/theme.
> The skinning model is different in FlexJS anyway so I think that gives us
> license to change the styling model as well.  If you ever set a style on a
> Flex component that wasn't in your custom theme, or tried to add a new
> style to a skin/theme that wasn't declared on the component, then you know
> what pain I'm talking about.  With FlexJS, some new browser might come up
> with a new custom CSS property and you won't be able to set it in MXML
> without us rev-ing the entire SDK.  Yes, you should be able to set it in
> the .CSS file, but just not in-line.
>
> Thoughts?
> -Alex
>
>
>

Re: [FlexJS] General questions

[Alex Harui]

On 9/8/13 10:01 PM, "flexcapaci...@gmail.com" 
wrote:

>Yes, I agree. What might alleviate that pain would be:
>
>€ move all the styles metadata into the skin instead of the component.
>have
>the component declare skin styles like we declare skin parts, so
>[SkinStyle(name="color", required=true)]
>€ or allow styles to be declared in the skin and make the compiler add
>those styles to the list of styles available to that component type for
>code completion and compilation
>€ or add a compiler flag to ignore attributes it is not aware of
>(highlight
>in a different color?) and treat those unknown attributes as styles.
>€ or add a styles proxy object on the component for styles not known to
>the
>component
>
>
>   
>
>
>I would prefer the above to inline, "color:red;cornerRadius:8".
>
>The more I think about it MXML should be a generic XML description of an
>application. Since we want MXML to output to other languages things like
>binding expressions and inline name value pairs like,
>style="fontFamily:Arial;fontWeight:bold" are getting away from the one of
>the reasons that XML was created; that being readability.
>
>I think another feature of an XML spec is that it defines guidelines for
>parsers. E4X is an advanced notation to get right to the information you
>need in an XML document. With, style="color:yellow;fontFamily:Times New
>Roman" you would need to use E4X to get the attribute and then another
>expression engine, RegExp, to get the names and values you need.
>
>I'm not saying no to inline styles but to keep in mind MXML's roots. In
>time, as MXML is adopted, the tooling and the developer base will grow as
>well and I would think we'd want a distributable document to be as
>portable
>and parsable as possible (say that 3 times fast).
All good thoughts and suggestions.   I'm still not sure what we're going
to do, but I've ruled out only using html-like inline styles for now.
>
>Are you using the style property as a proxy to the CSS class now?
>
>MXML,
>
>
>
>HTML,
>
>
>
Yes, and some properties like width and height get set in HTML on the
styles object.  Although today, it only works for known CSS properties.
I'm exploring what to do for custom CSS properties, especially in IE8.

The goal is to find best practices in HTML/JS then wrap and present them
in AS3.

-Alex

Re: Jenkins build is back to normal : flex-sdk_mustella-mobile #129

[Erik de Bruin]

Yes!

Thanks all for getting Mustella back to 'zarro boogs found' ;-)

As a bonus: the SDK (main) tests pass for 11.9/3.9!

EdB



On Mon, Sep 9, 2013 at 4:13 AM,   wrote:
> See 
>



-- 
Ix Multimedia Software

Jan Luykenstraat 27
3521 VB Utrecht

T. 06-51952295
I. www.ixsoftware.nl