You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@flex.apache.org by Joe Kryzak <fl...@aligned.com> on 2013/02/19 02:21:19 UTC
Worker/Mutex/Condition
Our applications run with Flex 4.6, we're considering upgrading to 4.7 for the concurrency features. Ideally, we would like to jump onto the apache flex bandwagon. In the published API I don't see Worker/Mutex/Condition objects - are these included in the latest release but the documents are behind? Or was there a decision made to let these classes go and thus apache flex does not support concurrency?
Thanks!
Joe
Re: Worker/Mutex/Condition
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi,
> A beginner question - in regard to Terry's point about the layering of the SDKs over the last available Flex SDK from Adobe, I'm not quite clear how additional capabilities in flash player are added to the apache flex development environment.
Currently Apache Flex does use any of the new capabilities added to the newer version of the Flash Player as a) we aim to be backwards compatible and b) I'm not sure any recents features added would improved the SDK in a significant way. If you know of any or have any ideas please suggest them!
The SDK can be compiled against newer versions of the SDK or you can use newer versions of the Flash Player or AIR fairly easily. Several scripts have been written to help you do this so you don't have the patch the SDK by hand. These are in the develop branch of the SDK under the ide folder and are likely to make it into the 4.10 release. Generally adding support for a new version of Flash/AIR just means changing the minimum FP version and SWF versions in the flex and air config files and downloading the latest playerglobal.swc.
> For instance, the Worker class is included in flash player 11.4, and Adobe Flex 4.7 has full support of flash player 11.4. As the Apache Flex SDK is overlaid on 4.7, how does a future flash player update, such as an addition of a Worker1 class, become accessible to an apache flex developer?
You would need to update the SDK to use this version of the Flash Player by editing the config files and downloading the latest playergloabl.swc.
> To Terry's other point, it seems that he is implying that there isn't complete accountability in testing for features. Is anyone working on test reports, or requirement driven verification?
Each release of Apache Flex will be tested with one or more versions of the Flash Player. Apache Flex 4.9.0 was tested on FP 11.1 and 11.4 and AIR 3.4. Apache Flex 4.10 will likely be tested with the latest version of Flash Player and AIR at the time of it's release. Currently we don't have the resources to test every release against every version of the Flash Player or make a new release whenever Adobe releases a new Flash Player/version of AIR.
I don't know if Adobe tests Adobe Flex 4.6 against newer versions of the Flash Player when they release them.
That being said it's very likely that Apache Flex 4.9 works with fine FP 11.6 and AIR 3.6 and support for both have already been added into the development branch.
> Is there a quantitative measure of confidence, ie code coverage, 97% passing, etc.
Testing is done by running the Mustella test suite. It work by compiling and running swf files, interacting with them, taking screen shots and comparing them with a baseline run. I don't know what code coverage of the SDK it tests (lines or paths) but it is fairly comprehensive. The full set of tests take 6-8 hours to run and contains 14,000+ tests (although I'm not sure of exact number).
There is a smaller group of check in tests and these are run via Jenkins on a CI machine when there's a check in to the develop branch.
You can read more about Mustela here:
https://cwiki.apache.org/FLEX/mustella-overview.html
Thanks,
Justin
Re: Worker/Mutex/Condition
Posted by Frédéric THOMAS <we...@hotmail.com>.
Joe,
Ok, you'll be more than wellcome if you want to contribute ;-)
-Fred
-----Message d'origine-----
From: Joe Kryzak
Sent: Tuesday, February 19, 2013 10:44 AM
To: Frédéric THOMAS
Subject: Re: Worker/Mutex/Condition
That makes sense. Thanks again, looking forward to contributing - we have a
busy few months ahead of us, and then it will slow down. In looking through
the bugs I find them fairly complex and over my head. I'm hoping once I
start diving in, it will become clearer.
Joe
On Feb 19, 2013, at 1:41 AM, Frédéric THOMAS wrote:
Joe,
To use the FP 11.6 specific features with Apache Flex SDK, simply declare it
in the flex config file and download the corresponding playerglobal.swc,
that's what he meant, that worths as well for AIR.
-Fred
-----Message d'origine----- From: Joe Kryzak
Sent: Tuesday, February 19, 2013 10:25 AM
To: Frédéric THOMAS
Subject: Re: Worker/Mutex/Condition
Thanks Fred for the response, it clarified my understanding. I thought
Justin was saying that new flash features weren't a capability.
I never thought of comparing bitmaps for our tests. Mostly right now I'm
using directed tests with requirements. My background is in aviation flight
control verification and semiconductors.
Joe
On Feb 19, 2013, at 1:01 AM, Frédéric THOMAS wrote:
Hi,
As Justin said, most of the tests are done comparing bitmaps, that's more
functional testing than unit testing but there are also method, property and
error tests.
It's difficult to say the percentage of code coverage we've got, adobe has a
tool doing that but I don't know the status of its donation but knows that
it's not 14000+ tests we've got, it's 36000+ tests, theoricaly, they all
should pass on mac and windows with Flash Player 10.2 to 11.6+ but it hasn't
been fully tested on each ones as stated before, what I can say is that I
ran the test suite few days ago and on my windows 7, I had only 1 test
failed and it seems to pass on mac, for the 4.9.0 I tested it against FP
11.5 and had 4 tests failed which means praticly, we're > 99.988 % of passed
tests on players from FP11.1 to 11.5 and those failed tests are only few
pixels rendering differences, nothing catastrophic and as you can see,
unlike Adobe SDKs, we support more Flash Players, it means that you can
leverage users with low requirements regarding the FP version they use or
leverage the newest flash player features.
-Fred
-----Message d'origine----- From: Justin Mclean
Sent: Tuesday, February 19, 2013 7:47 AM
To: users@flex.apache.org
Subject: Re: Worker/Mutex/Condition
Hi,
> Third, do you guys recommend we go with 4.7 for a production update, or do
> you feel confident in the latest releases? Is there a quantitative measure
> of confidence, ie code coverage, 97% passing, etc.
Forgot to say policy is all test must pass for it to be an official release.
Justin
Re: Worker/Mutex/Condition
Posted by Frédéric THOMAS <we...@hotmail.com>.
Hi,
As Justin said, most of the tests are done comparing bitmaps, that's more
functional testing than unit testing but there are also method, property and
error tests.
It's difficult to say the percentage of code coverage we've got, adobe has a
tool doing that but I don't know the status of its donation but knows that
it's not 14000+ tests we've got, it's 36000+ tests, theoricaly, they all
should pass on mac and windows with Flash Player 10.2 to 11.6+ but it hasn't
been fully tested on each ones as stated before, what I can say is that I
ran the test suite few days ago and on my windows 7, I had only 1 test
failed and it seems to pass on mac, for the 4.9.0 I tested it against FP
11.5 and had 4 tests failed which means praticly, we're > 99.988 % of passed
tests on players from FP11.1 to 11.5 and those failed tests are only few
pixels rendering differences, nothing catastrophic and as you can see,
unlike Adobe SDKs, we support more Flash Players, it means that you can
leverage users with low requirements regarding the FP version they use or
leverage the newest flash player features.
-Fred
-----Message d'origine-----
From: Justin Mclean
Sent: Tuesday, February 19, 2013 7:47 AM
To: users@flex.apache.org
Subject: Re: Worker/Mutex/Condition
Hi,
> Third, do you guys recommend we go with 4.7 for a production update, or do
> you feel confident in the latest releases? Is there a quantitative measure
> of confidence, ie code coverage, 97% passing, etc.
Forgot to say policy is all test must pass for it to be an official release.
Justin
Re: Worker/Mutex/Condition
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi,
> Third, do you guys recommend we go with 4.7 for a production update, or do you feel confident in the latest releases? Is there a quantitative measure of confidence, ie code coverage, 97% passing, etc.
Forgot to say policy is all test must pass for it to be an official release.
Justin
Re: Worker/Mutex/Condition
Posted by Joe Kryzak <fl...@aligned.com>.
A beginner question - in regard to Terry's point about the layering of the SDKs over the last available Flex SDK from Adobe, I'm not quite clear how additional capabilities in flash player are added to the apache flex development environment. For instance, the Worker class is included in flash player 11.4, and Adobe Flex 4.7 has full support of flash player 11.4. As the Apache Flex SDK is overlaid on 4.7, how does a future flash player update, such as an addition of a Worker1 class, become accessible to an apache flex developer?
To Terry's other point, it seems that he is implying that there isn't complete accountability in testing for features. Is anyone working on test reports, or requirement driven verification?
Third, do you guys recommend we go with 4.7 for a production update, or do you feel confident in the latest releases? Is there a quantitative measure of confidence, ie code coverage, 97% passing, etc.
Thanks, looking forward to getting more involved.
On Feb 18, 2013, at 6:38 PM, Justin Mclean wrote:
Hi,
> In the time since, I, like I presume most, have been waiting to see whether or not we can get back to that state of affairs using the Apache Flex SDK with the most current AIR SDK layered from Adobe. Our current understanding is that yes, we can 'sort of try it', but because the new AAC2 Compiler cannot handle our Spark Componentry when injected via MXML, we really are not yet back to square one. As far as I can tell, the same ambiguity now exists with regards to Workers.
As far as I'm aware the new compiler is not required to use AS workers. AIR 3.4 and 3.5 work well with Apache Flex and I tried out AIR 3.6 the other day (nothing complex however). What exactly is your problem with AIR 3.6 and ApacheFlex?
Thanks,
Justin
Re: Worker/Mutex/Condition
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi,
> In the time since, I, like I presume most, have been waiting to see whether or not we can get back to that state of affairs using the Apache Flex SDK with the most current AIR SDK layered from Adobe. Our current understanding is that yes, we can 'sort of try it', but because the new AAC2 Compiler cannot handle our Spark Componentry when injected via MXML, we really are not yet back to square one. As far as I can tell, the same ambiguity now exists with regards to Workers.
As far as I'm aware the new compiler is not required to use AS workers. AIR 3.4 and 3.5 work well with Apache Flex and I tried out AIR 3.6 the other day (nothing complex however). What exactly is your problem with AIR 3.6 and ApacheFlex?
Thanks,
Justin
Re: Worker/Mutex/Condition
Posted by Terry Corbet <tc...@ix.netcom.com>.
I suppose this is an answer to the OP question, but it does not seem so to
me.
When workers were introduced/tested/released, I went through the Beta cycle
and subsequently devoloped multi-worker AIR applications layering the AIR
SDK atop the last available Flex SDK from Adobe.
In the time since, I, like I presume most, have been waiting to see whether
or not we can get back to that state of affairs using the Apache Flex SDK
with the most current AIR SDK layered from Adobe. Our current understanding
is that yes, we can 'sort of try it', but because the new AAC2 Compiler
cannot handle our Spark Componentry when injected via MXML, we really are
not yet back to square one. As far as I can tell, the same ambiguity now
exists with regards to Workers.
Can someone post an official, unambiguous answer to where these key
capabilities stand?
It is painfully clear that the Gaming Product Manager for all things
Actionscript is doing his best to distance his team's efforts from those of
the Flex community. The Spoon folks wanted this challenge; will the Spoon
and/or Adobe folks please stand up and give us their answer to where things
stand especially since the release of the last, verbose, unhelpful Statement
of Direction from Adobe since the killing off of ActionScript Next.
----- Original Message -----
From: "Justin Mclean" <ju...@classsoftware.com>
To: <us...@flex.apache.org>
Sent: February 18, 2013 06:07 PM
Subject: Re: Worker/Mutex/Condition
Hi,
> Our applications run with Flex 4.6, we're considering upgrading to 4.7 for
> the concurrency features.
The concurrency features are part of the Flash Player (first added in 11.4)
not the Adobe Flex SDK. The Adobe Flex SDK does support flash player 11.4
and does compile against 11.5 and 11.6 but hasn't been fully tested with
those versions.
Thanks,
Justin
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2013.0.2899 / Virus Database: 2639/6110 - Release Date: 02/17/13
Re: Worker/Mutex/Condition
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi,
> Our applications run with Flex 4.6, we're considering upgrading to 4.7 for the concurrency features.
The concurrency features are part of the Flash Player (first added in 11.4) not the Adobe Flex SDK. The Adobe Flex SDK does support flash player 11.4 and does compile against 11.5 and 11.6 but hasn't been fully tested with those versions.
Thanks,
Justin