Understood.
In this case, I' am changing the test code so that it matches the existing
baseline, not changing the baseline.
So it should not have any impact.
Maurice
-Message d'origine-
De : Alex Harui [mailto:aha...@adobe.com]
Envoyé : jeudi 10 octobre 2013 02:43
À : dev@flex.apache.
The test doesn't seem to be setting htmlText to null, but I haven't stepped
through it.
First, I want to figure out what went wrong with Git, then I'll try to find
time to investigate the original bug and proposed fix.
I'm done for tonight. More tomorrow...
-Alex
__
On Wed, Oct 9, 2013 at 8:07 AM, Alex Harui wrote:
> I checked in a bunch of fixes to Falcon to get a large 20MB SWF to compile
> and get to its first screen. I'm sure there are more bugs to fix if I used
> that SWF, but I'm going to stop there for a bit and work on other issues.
I can confirm
https://issues.apache.org/jira/browse/FLEX-33307
The test is currently @Ignore, but there are I think about 3 (types
of) errors when you run it. It is the only test blocking Falcon
compiling all the SDK SWCs. A couple of other tests generate warnings,
that are currently '-ignore-problem', that mig
Here is the commit history to the file:
https://fisheye6.atlassian.com/browse/~br=develop/flex-sdk/frameworks/projects/spark/src/mx/core/FTETextField.as
I dont see Nick's revert.
Thanks,
Om
On Wed, Oct 9, 2013 at 11:07 PM, Justin Mclean wrote:
> Hi,
>
> > When I look at the diffs, I don't see
Hi,
> When I look at the diffs, I don't see the clearFlag line until Nick adds it.
> The comment above the "if (!stylesheet)" makes me think the flag shouldn't be
> cleared, but I haven't really looked into it.
If htmlText is set to null I think the HTML flag needs to be cleared, I think
(but
When I look at the diffs, I don't see the clearFlag line until Nick adds it.
The comment above the "if (!stylesheet)" makes me think the flag shouldn't be
cleared, but I haven't really looked into it.
I also took out the clearFlag call and the tests passed.
Right now, I'm more puzzled why Ni
Hi,
> What do you see that's wrong with the tests?
If you look at the fix:
From:
if (!styleSheet)
htmlText = null;
textfield.clearFlag(FTETextField.FLAG_HTML_TEXT_SET);
To:
if (!styleSheet)
{
htmlText = null;
textfield.clearFlag(F
Update: These same two tests fail for me on OSX.
My investigation has discovered that after a "git pull" on both my Windows and
Mac, neither show the revert of Nick's change to FTETextField, and the git log
doesn't show it either. What do you guys see when you look at
frameworks/projects/spar
On 10/10/2013 2:52 AM, Michael A. Labriola wrote:
Icu4j (actually icu4c) is what we used as the basis for all of our work too.
It has a ton of great stuff in there. In particular we were also dealing with
non-gregorian calendaring and they have some reasonably helpful code
throughout.
i have a
It would be interesting to run the same test on machines with a different
number of processors. In theory, Falcon should outperform MXMLC as the
number of processors increase.
On Wed, Oct 9, 2013 at 11:38 PM, Justin Mclean wrote:
> Hi,
>
> > No official measurements. How small are you talking a
Hi,
> No official measurements. How small are you talking about?
Was just wondering if the performance gain were hampered by disk access or
other bottle necks, a smaller project may give an more reasonable idea of the
typical performance gain in using Falcon. What exactly is a typical project
Update: I synced up and ran the FTETextField tests on my Windows box and got
the same 2 errors (FP11.1).
I'll be doing some digging... Won't need the Mustella VM.
-Alex
From: flex.muste...@gmail.com [flex.muste...@gmail.com]
Sent: Tuesday, October 08, 2
I thought there were a few open issues. IIRC, a bug I have assigned to me was
related. It is time to try again and see if anything is still broken.
-Alex
From: Justin Mclean [jus...@classsoftware.com]
Sent: Wednesday, October 09, 2013 6:09 PM
To: dev@fle
No official measurements. How small are you talking about?
From: Justin Mclean [jus...@classsoftware.com]
Sent: Wednesday, October 09, 2013 5:56 PM
To: dev@flex.apache.org
Subject: Re: [FALCON] Large App Compiles and Starts
Hi,
> The Ant script that build
Hi,
> The next milestone is to get the Flex SDK to compile with Falcon. Any bugs
> related to that would be good to work on.
This has been marked as done in JIRA - is there something that still needs to
be done?
Thanks,
Justin
Hi,
> The Ant script that builds all of this code (several SWCs then one big SWF)
> takes about 4 minutes to run using MXMLC and 3.5 minutes with Falcon.
Do you have some ballpark performance gains re Falcon vs MXMLC and smaller swfs?
Thanks,
Justin
When you change a test please consider impact on folks who might have pixel
tuned their app and may notice the difference. I'm not saying don't change, but
be careful.
Sent via the PANTECH Discover, an AT&T 4G LTE smartphone.
Maurice Amsellem wrote:
No, because the test on line 579 is:
va
Will do that tomorrow.
Have to go to bed now.
Maurice
-Message d'origine-
De : Justin Mclean [mailto:jus...@classsoftware.com]
Envoyé : jeudi 10 octobre 2013 01:48
À : dev@flex.apache.org
Objet : Re: Build failed in Jenkins: flex-sdk_mustella #460 (fixed)
Hi,
> Should I change the tes
Hi,
> Should I change the test ?
Seems the right option to me, we're changed the code slightly we can change the
test to match.
Thanks,
Justin
Ok, finally found out for the last failure:
Actually, one of the tests in using a custom callout skin,
SplitViewNavigator/swfs/skins/MyCalloutSkin.as which constructor is the
following:
public function MyCalloutSkin()
{
super();
No, because the test on line 579 is:
var showBorder:Boolean = !isNaN(borderThickness);
In the desktop skin, I am setting it that way:
var showBorder:Boolean = !isNaN(borderThickness) || borderThickness > 0 ;
Should I change the test ?
Maurice
-Message d'origine-
De : Justin Mclean
Hi,
> So the variable showBorder on line 579 is set to true.
If borderThinkness is 0 shouldn't showBorder be false? Then the callout should
be in the right position.
Thanks,
Justin
Option 1) removing the added lines, eliminates all the failures.
While Option 2) [ adding the missing borderThickness:NaN ] stills has one
failure.
So if you think the border option will be useful in mobile skins as well (
maybe on Android...) I can work on removing the last failure.
Otherwis
I found out:
I have added "borderStyle" and "borderThickness" styles to Callout class, so
that desktop skin have a 1pix border and look more like other components skins.
For consistency, I also changed the Callout and CalloutArrow mobile skins code
as well, so that it uses these styles:
411
You are correct.
>From the "bad.png" files (eg. Callout_skin@android_240ppi.png.bad.png) , I
>can see it's using "android" extension for me.
Maurice
-Message d'origine-
De : Justin Mclean [mailto:jus...@classsoftware.com]
Envoyé : jeudi 10 octobre 2013 00:52
À : dev@flex.apache.org
Hi,
> I think when it's running on ADL on windows , it will use "win". Correct ?
I assumed it would use the android one, if emulating android it should be
comparing the android images right?
On OSX.
With this local.properties:
target_os_name=android
device_name=osx
And running the tests like
I found out at list one difference:
In change_skin_at_runtime, the callout in exactly the same, but shifted one
pixel to the right.
Will try some changes in the mobile callout skin to see if the difference
disappears.
Will let you know.
-Message d'origine-
De : Maurice Amsellem [mail
-createImages will only create new baselines which should pass because they are
coming from the same computer. I don't think that will tell you anything.
You have to look at the CompareBitmap in the test to see which bitmap it will
use. The default doesn't have any device name in the file path
Finally, the .bad.png images are also there for the 6 initial 240 ppi failures
in the baseline dirs.
YES!
Comparing them now...
Maurice
-Message d'origine-
De : Maurice Amsellem [mailto:maurice.amsel...@systar.com]
Envoyé : jeudi 10 octobre 2013 00:36
À : dev@flex.apache.org
Objet : RE
Hi,
> Note: the error says " created image .bad.png", does it also create these
> images when baseline is found, but compare fails ?
> If so, where do I find it ?
In the baselines directory for the test.
Justin
I will give a try to the createBitmaps option.
BTW it will show what bitmaps are actually used...
Maurice
-Message d'origine-
De : Maurice Amsellem [mailto:maurice.amsel...@systar.com]
Envoyé : jeudi 10 octobre 2013 00:33
À : dev@flex.apache.org
Objet : RE: Build failed in Jenkins: fle
I found the .bad.png images, they are in the same baselines directory.
Maurice
-Message d'origine-
De : Maurice Amsellem [mailto:maurice.amsel...@systar.com]
Envoyé : jeudi 10 octobre 2013 00:30
À : dev@flex.apache.org
Objet : RE: Build failed in Jenkins: flex-sdk_mustella #460
I have
But maybe it's using the "android" bitmaps, like you said, not the "win" ones.
Maurice
-Message d'origine-
De : Justin Mclean [mailto:jus...@classsoftware.com]
Envoyé : jeudi 10 octobre 2013 00:28
À : dev@flex.apache.org
Objet : Re: Build failed in Jenkins: flex-sdk_mustella #460
Hi,
I have made another run using
adl_extras=-screensize 320x455:320x480 -profile mobileDevice -XscreenDPI 160
so that it uses the 160 dpis baselines.
I am getting completely different other errors:
[java] mobile/SplitViewNavigator/integration/SplitViewNavigator_SPUC
PopUp_open Failed CompareBi
Hi,
> If I backup the baselines dir and run the test with -createImages option, I
> will be able to check the differences.
Bit hard to check the differences if they don't exist in the first place. :-)
Justin
I also have target_os_name=android, and device_name=win.
I think when it's running on ADL on windows , it will use "win". Correct ?
Maurice
-Message d'origine-
De : Justin Mclean [mailto:jus...@classsoftware.com]
Envoyé : jeudi 10 octobre 2013 00:24
À : dev@flex.apache.org
Objet : Re:
If I backup the baselines dir and run the test with -createImages option, I
will be able to check the differences.
Right ?
Maurice
-Message d'origine-
De : omup...@gmail.com [mailto:omup...@gmail.com] De la part de OmPrakash
Muppirala
Envoyé : jeudi 10 octobre 2013 00:20
À : dev@flex.a
Hi,
> integration\baselines\Skin_test_1@win_240ppi.png
> integration\baselines\Change_skin_at_runtime@win_240ppi.png
> integration\baselines\Callout_skin @win_240ppi.png
I think it should be using the @android bitmaps. I have target_os_name=android
in local.properties (as mentioned before) - per
>From the VM's failures.txt, these are the failing tests:
mobile/SplitViewNavigator/integration/SplitViewNavigator_integration
Skin_test_1
mobile/SplitViewNavigator/integration/SplitViewNavigator_integration
Change_skin_at_runtime
mobile/SplitViewNavigator/integration/SplitViewNavigator_integratio
See my next email about missing baseline bitmaps.
I am running on windows (so set device-name to win)
-Message d'origine-
De : Justin Mclean [mailto:jus...@classsoftware.com]
Envoyé : jeudi 10 octobre 2013 00:16
À : dev@flex.apache.org
Objet : Re: Build failed in Jenkins: flex-sdk_mustel
Hi,
> Is it possible that it's related to the Callout merge ?
Those tests do use callouts so it possible but may not be the issue.
How are you funning the tests? Running here I get no failures and all 50 odd
tests pass.
Thanks,
Justin
I have looked for the different baseline bitmaps to compare:
integration\baselines\Skin_test_1@win_240ppi.png
integration\baselines\Change_skin_at_runtime@win_240ppi.png
integration\baselines\Callout_skin @win_240ppi.png
etc.
And the bitmaps are all missing from
mustella\tests\mobile\SplitViewN
Hi, I have run the SplitNavigatorView and also had the 6 fails:
[java] mobile/SplitViewNavigator/integration/SplitViewNavigator_integration
Skin_test_1 Failed CompareBitmap(body:step 4) compare returned-3
[java]
mobile/SplitViewNavigator/integration/SplitViewNavigator_integration
Chan
The next milestone is to get the Flex SDK to compile with Falcon. Any bugs
related to that would be good to work on.
-Alex
From: Darrell Loverin [darrell.love...@gmail.com]
Sent: Wednesday, October 09, 2013 10:22 AM
To: dev@flex.apache.org
Subject: Re: [F
> For Doug and Mike, was it really easier to do custom AMF and timezone math
> vs just creating a custom DateWithoutTime class and transfer the Date
> portion to/from the Date values in the components?
Note that my requirement was that we did indeed need the time element, we
just always wanted it
>I will have a look at icu4j.
Icu4j (actually icu4c) is what we used as the basis for all of our work too. It
has a ton of great stuff in there. In particular we were also dealing with
non-gregorian calendaring and they have some reasonably helpful code throughout.
We ended up taking the transi
I will have a look at icu4j.
Thanks
Maurice
__
De : Paul Hastings [paul.hasti...@gmail.com]
Envoyé : mercredi 9 octobre 2013 21:23
À : Maurice Amsellem
Cc : dev@flex.apache.org
Objet : Re: Date and Time with TZ discussion
On 10/9/2013 7:16 PM, Maurice Amse
I am trying to build out a Dojo based component set (like jquery, createjs,
etc.) Which is why getting the jquery based app running is crucial for
me.
Are you sure you haven't made changes to the example code?
Or, perhaps you publish a new overlay zip? I am not sure if you have been
pushing new
On 10/9/2013 7:16 PM, Maurice Amsellem wrote:
This is the approach I have taken to solve this, that successfully addresses
my requirements (monitoring apps for world-wide Banks with offices in
different continents). I don't know if it makes sense to other usages, but I
think you could be interes
Om,
It ran for me, so I don't know what else to fix. I'm thinking you have some
sort of config problem. What does the console output look like? Maybe there's
a clue in there.
Peter and I are using the DataBindingTest app that doesn't use JQuery. I'm not
sure what your goal is and whether J
On 10/9/2013 9:23 PM, Michael A. Labriola wrote:
for "historical" dates we've used calls back to the server (coldfusion) or
during data retrieval where we could access the whole tz database. but the
issue then became what's >"historical"? which ended up as "whatever the
client said it was", not t
Bump...
Alex/Peter, if you can take some time to fix the jquery example app that
would be great. Or, if you want me to switch to any other app for
reference, please advice.
Thanks,
Om
On Thu, Oct 3, 2013 at 4:01 PM, OmPrakash Muppirala wrote:
> On Wed, Oct 2, 2013 at 11:58 PM, Alex Harui wro
Hi,
I've spent some time on an experiment in FlexJS. Right now we have all of the
components sitting in one package called, "staticControls". The intent of this
package was to have controls that were relatively simple and had direct or
close approximation with HTML elements. For example, Button
I was using it while I was a customer that purchased the BlazeDS product.
I haven't since the 2.6.1 days
-Nick
On Wed, Oct 9, 2013 at 12:10 PM, Alex Harui wrote:
> To be clear, this is not about the source for the 1.1 compiler or any
> compiler that supports AS2. Apparently there is a co
Ah sorry.
That what happens when you don't read the whole thread :-).
Maurice
-Message d'origine-
De : omup...@gmail.com [mailto:omup...@gmail.com] De la part de OmPrakash
Muppirala
Envoyé : mercredi 9 octobre 2013 20:01
À : dev@flex.apache.org
Objet : Re: Build failed in Jenkins: flex-
On Wed, Oct 9, 2013 at 10:58 AM, Maurice Amsellem <
maurice.amsel...@systar.com> wrote:
> Hi, when browsing the history on develop branch, last change on
> SplitViewNavigator is back from 4/26/2013 and it's really a minor change.
> Did I miss something ?
>
>
Alex mentioned that it could be related
Hi, when browsing the history on develop branch, last change on
SplitViewNavigator is back from 4/26/2013 and it's really a minor change.
Did I miss something ?
Maurice
-Message d'origine-
De : Erik de Bruin [mailto:e...@ixsoftware.nl]
Envoyé : mercredi 9 octobre 2013 19:33
À : dev@fle
Develop branch, of course.
And the current status is that both 'main' and 'mobile' have failing
tests. Where the 'main' failures seem mostly related to FTETextField
and the failures for 'mobile' center around SplitViewNavigator.
The VM runs on Windows Server 2008. The three Mustella test suites r
Nice job Alex! Let me know if there is a bug you would like me to pick up
while you work on something else.
On Wed, Oct 9, 2013 at 1:09 PM, Gordon Smith wrote:
> Congratulatlons on reaching a huge milestone!
>
> - Gordon
>
> -Original Message-
> From: Alex Harui [mailto:aha...@adobe.com
Congratulatlons on reaching a huge milestone!
- Gordon
-Original Message-
From: Alex Harui [mailto:aha...@adobe.com]
Sent: Tuesday, October 08, 2013 11:07 PM
To: dev@flex.apache.org
Subject: [FALCON] Large App Compiles and Starts
I checked in a bunch of fixes to Falcon to get a large 20
I will try to dig around this evening (9pm west coast) when I can get to my
Windows computer.
From: Nicholas Kwiatkowski [nicho...@spoon.as]
Sent: Wednesday, October 09, 2013 8:48 AM
To: dev@flex.apache.org
Subject: Re: Build failed in Jenkins: flex-sdk_mus
The fonts should be hard coded. You'd probably get more pixel differences if
not. It is just that sometimes the player chooses a different value in the
anti-aliasing. We can create and check in a png.xml file for the test that is
failing and that should cause it to pass on Linux.
I highly suspect anyone would be using web-tier compiler anymore, in the
remotest possibility that anyone is
It can be easily replaced by a java wrapper script that invokes mxmlc
Thanks,
Pratyoosh
-Original Message-
From: Alex Harui [mailto:aha...@adobe.com]
Sent: Wednesday, October 09, 2
To be clear, this is not about the source for the 1.1 compiler or any compiler
that supports AS2. Apparently there is a compiler that supports Flex 3 and
AS3. Is anyone using that?
-Alex
From: Nicholas Kwiatkowski [nicho...@spoon.as]
Sent: Wednesday, Oc
Because CF still uses the 1.1 Flex compiler... Which didn't have a
standalone compiler..
Better yet, there is now about 8 years of crazy as2 code out there built in
CF that depends on this old version.
Sigh.
On Wed, Oct 9, 2013 at 4:06 AM, Tom Chiverton wrote:
> On 09/10/2013 02:16, Nick Col
That would be good to know... I could never get the FTETextField tests to
fail on my own copy of Mustella. Seems odd that the tests passed once then
failed again later...
On Wed, Oct 9, 2013 at 2:37 AM, Justin Mclean wrote:
> Hi,
>
> Is the testing VM using the master or develop branch of TLF?
>for "historical" dates we've used calls back to the server (coldfusion) or
>during data retrieval where we could access the whole tz database. but the
>issue then became what's >"historical"? which ended up as "whatever the client
>said it was", not the best approach.
Problem with the server c
>If the patch is small, they should just submit the patch in a JIRA and an ICLA
>is not required.
It will eventually come in via jira but there are other hurdles to pass first.
It is using work from others, likely binaries and a ton of other issues that
all need to be resolved before the patch
On 09/10/2013 13:17, Paul Hastings wrote:
Well I enjoy backwards compatibility,
It's not like CFN just stops working when CFN+1 comes out, you can keep
it around for as long as you like...
But we may be straying off topic ;-)
Tom
Well I enjoy backwards compatibility, so I guess have to put up w/the
occasional 'blip'.
Paul Hastings said:
> really curious to see what folks have done to solve this.
This is the approach I have taken to solve this, that successfully addresses my
requirements (monitoring apps for world-wide Banks with offices in different
continents).
I don't know if it makes sense to other usages
On 09/10/2013 12:13, Paul Hastings wrote:
On 10/9/2013 3:06 PM, Tom Chiverton wrote:
using the compile-on-the-fly one. I have no idea why things like
ColdFusion still ship it !
flash forms.
Like I said, I have no idea why ColdFusion ships these things :-)
Tom
On 10/9/2013 3:06 PM, Tom Chiverton wrote:
using the compile-on-the-fly one. I have no idea why things like
ColdFusion still ship it !
flash forms.
Hi,
> does the automated testing machine check on Linux at all ?
Not for a long time (if ever) is my guess.
Justin
Hi list,
I just noticed the commit to fix the AdvancedDataGrid issue:
FLEX-32728 (After selectedIndices was updated, it can not be selected with
ShiftKey).
Some days ago I opened the Spark DataGrid issue:
FLEX-33774 (grid.anchorRowIndex is reset to 0 when selection is
programatically set (breakin
On 09/10/2013 09:12, Justin Mclean wrote:
Hi,
I get a single test failure (pixel color wrong), both before and after my patch
- I'll update ExcludeListLin.txt later.
For a single pixel out I'd not exclude the test and it may just be OS/FP/font
difference.
I'll try and dig out the bad/good
Hi,
> I get a single test failure (pixel color wrong), both before and after my
> patch - I'll update ExcludeListLin.txt later.
For a single pixel out I'd not exclude the test and it may just be OS/FP/font
difference.
Justin
On 09/10/2013 02:16, Nick Collins wrote:
I'm assuming the web tier compiler is the one where we can load an MXML
file from the server much like one would load a JSP and it will compile the
app on first request and return the SWF? I personally don't have a need for
that, but others may. I'd rather
On 08/10/2013 22:44, Justin Mclean wrote:
On OSX this passes for me:
./mini_run.sh tests/gumbo/components/DataGrid
Cheers, I did get it to run now, only crashing Xorg three times in the
process :-)
I get a single test failure (pixel color wrong), both before and after
my patch - I'll update
Justin Mclean said:
>And don't forgot the daylight savings time "feature", this comes up again and
>again in Adobe's JIRA over the years.
That's what I called "manage DST properly".
Yes of course, it needs to be managed, otherwise it's not funny :-).
Maurice
-Message d'origine-
De : M
(Moved this topic on itw own thread and summed up the answers so far, so that
we can continue the discussion)
Maurice
-Message d'origine-
De : Paul Hastings [mailto:paul.hasti...@gmail.com]
Envoyé : mercredi 9 octobre 2013 06:34
À : dev@flex.apache.org
Objet : Re: Date without Time (Wa
Let me elaborate on my previous comment.
In my opinion, we could introduce a new type (or more than one), but we
should make modifications to framework components.
For example, the DateField may use the new DateWithoutTime class directly,
because it doesn't supports time resolution in user intera
83 matches
Mail list logo