Something new to worry about

[Michael Labriola]

With FlashPlayer 10.1, when the window containing the browser running
tests no longer has focus, the framerate is slowed. This is an
optimization for mobile, but one that is about to cause us a world of
hurt.

If we run tests in the background now, all of the async tests fail as
the framerate is so slow, there is little chance of success. I haven't
been able to find a way to turn this off for our purposes and I am
expecting we will see more issues here as users continue to adopt
10.1. The only thing I have been able to think of as a possible
workaround (which has its own problems) would be to adapt all timeouts
to framerate.

In other words, not keep the number of milliseconds something is
supposed to take, but rather look at those milliseconds as a function
of normal framerate and adapt as the framerate changes. I already know
a handful of issues with this approach, but I would like some thoughts
on all of the ones I am missing or some alternate idea brainstorming.
Anything would help as I am sure we will begin to hear about this
soon.

Mike

[Aaron Boushley]

I don't remember all the details, and I can't find any articles on it
right now, but I know that the new player also treats Flash Player
instances that are only 1pixel x 1pixel in a special manner, assuming
that they are there for non rendering purposes. I can't remember off
the top of my head if these special behaviors had to do with only when
the instances are loaded or if it had to do with the framerate as well
( I also don't remember if these special cases are only for mobile ).

So I'm not sure if this will be useful to us, but if these are exempt
from the adaptive framerate then we may be able to look into using a
1x1 instance to run the tests. Again, not sure about all of these
things I'll do a little more research and post again, but just a
thought to kick around.

Aaron

[Borek Bernard]

Mike,

this is what first came to mind: FlexUnit is now part of a supported
commercial IDE provided by Adobe and if Adobe changes their related
technology in a way that it "breaks" something in Flash Builder, it
should be their responsibility to "fix" it or find a workaround. The
frame rate thing is potentially a big problem for unit testing in the
Flash world and IMO, Adobe should be deeply concerned about it and
actively trying to fix it.

I know this comment is not technical, I just think that you have a
moral right to demand some solution from Adobe. If you create a ticket
on bugs.adobe.com the community will be able to vote on that but of
course if you have any relevant contacts in Adobe, that would be even
better.

Regards,
Borek

[Conrad Winchester]

Hi Guys,

I don't know if its of any help, but I Have set up a nexus repository for flash dependencies. The url is

http://www.flashnexus.org/nexus

Snazzy, huh?

I did this because I am moving towards maven as a build system and need to have loads of swc's in a nexus repository, and I am sure it would help other people too.

I had a response from Marvin Froeder when I asked him about getting a repo for Flexunit 4.1-beta2:

"Well, flexunit guys should provide a maven repo."

and so I thought I'd do it.

Unfortunately I am still learning about nexus and am not sure how to upload flexunit to it properly - Brian, do you know - if so do you want to have access to the repository and set it up?

Is this a helpful thing?

Conrad

P.S. I am having problems setting up the DNS correctly and so if you set your host file to point www.flashnexus.org to 212.13.203.133 it should all work for now.

[Brian]

@Conrad - I've been working on this slowly over the last few weeks.
We have a Sonatype repo, but have yet to publish anything to it since
I'm trying to use Apache Ivy and still comply with the Sonatype POM
guidelines. If I can work through it, then hopefully flexunit will
get replicated to Central as well. For now however, we don't have the
4.1 artifacts available in any type of artifact repository. I've been
keeping updates on my progress @ https://bugs.adobe.com/jira/browse/FXU-27.
If you want to ping me on IM, I can help you work through uploading to
your Nexus repo. I do like Nexus, but if you are running into issues
with it, check out Artifactory as well; I've heard good opinions about
both. I'll hit up the guys and ask him what he thinks about using
flashnexus vs Sonatype vs rolling our own instance.

I'm glad to see that others are starting to show interest in the
dependency management aspect of the build process, hopefully I can get
Ivy working for those folks who are still in Ant.

-Brian


On Aug 1, 11:21 am, Conrad Winchester

> P.S. I am having problems setting up the DNS correctly and so if you set your host file to pointwww.flashnexus.orgto 212.13.203.133 it should all work for now.

[Robert Penner]

A few months ago, I asked Tinic about a similar issue where Flash Player 10.1 limits the number of Timers, setIntervals, etc. He said there is a setting in mm.cfg that will remove the limitation for the Debug Player. Perhaps there is a setting for the background issue.

I haven't come across a list of mm.cfg variables up-to-date with 10.1, but here are good references for 10:

http://www.sephiroth.it/weblog/archives/2010/02/mmcfg_revealed.php

http://jpauclair.net/2010/02/10/mmcfg-treasure/

Robert

On Wed, Jul 28, 2010 at 8:49 AM, Michael Labriola <malab...@gmail.com> wrote:

[Michael Labriola]

Thanks guys. Trying to make some headway with adobe on this now

On Aug 3, 8:20 pm, Robert Penner <i...@robertpenner.com> wrote:
> A few months ago, I asked Tinic about a similar issue where Flash Player
> 10.1 limits the number of Timers, setIntervals, etc. He said there is a
> setting in mm.cfg that will remove the limitation for the Debug Player.
> Perhaps there is a setting for the background issue.
>
> I haven't come across a list of mm.cfg variables up-to-date with 10.1, but
> here are good references for 10:
>
> http://www.sephiroth.it/weblog/archives/2010/02/mmcfg_revealed.php
>
> http://jpauclair.net/2010/02/10/mmcfg-treasure/
>
> Robert
>