You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flex.apache.org by Kessler CTR Mark J <ma...@usmc.mil> on 2013/09/05 15:47:53 UTC
RE: Build failed in Jenkins: flex-sdk_mustella #290
Not sure how we changed a bitmap for variableRowHeight with the 3 below commits. I can look at the bitmaps when I get home. If it looks correct, how do I update the bitmap it uses to check with? Just copy it's compared one and push it back to the sdk?
-Mark
-----Original Message-----
From: flex.mustella@gmail.com [mailto:flex.mustella@gmail.com]
Sent: Thursday, September 05, 2013 9:30 AM
To: commits@flex.apache.org; KesslerConsulting@gmail.com; jmclean@apache.org; bigosmallm@gmail.com; fthomas@apache.org; cyrill.zadra@gmail.com
Subject: Build failed in Jenkins: flex-sdk_mustella #290
See <http://localhost:8080/job/flex-sdk_mustella/290/changes>
Changes:
[KesslerConsulting] FLEX-33710: Added index bounds checks.
[KesslerConsulting] FLEX-33710: Changed the Range out of bounds check to a try/catch for a range error.
[jmclean] 4.9 AIR support/11.9 Flash Player support
[java] =====================================================
[java] Failed:
[java] =====================================================
[java] gumbo/components/DataGrid/Properties/DataGrid_Properties_editable Editable_variableRowHeight Failed CompareBitmap(body:step 12) compare returned[object BitmapData]
[java]
[java]
[java] =====================================================
[java] Passes: 4
[java] Fails: 1
[java] =====================================================
[java]
[java]
[java] Wrote summary to results.txt
[java] Wrote failures to failures.txt
BUILD FAILED
C:\jenkins_slave\workspace\flex-sdk_mustella\mustella\build.xml:1538: The following error occurred while executing this line:
C:\jenkins_slave\workspace\flex-sdk_mustella\mustella\build.xml:1656: Java returned: 1
Total time: 1 minute 40 seconds
ute shell' marked build as failure
RE: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Kessler CTR Mark J <ma...@usmc.mil>.
Not sure how to tell all the settings its using, but the bottom of the Jenkins report says
Settings:
player.version = 11.1
air.version = 3.7
Hope fully it's something simple like you're anti-aliasing idea.
Thanks Alex,
-Mark
-----Original Message-----
From: Alex Harui [mailto:aharui@adobe.com]
Sent: Thursday, September 05, 2013 12:41 PM
To: dev@flex.apache.org
Subject: Re: Build failed in Jenkins: flex-sdk_mustella #290
On 9/5/13 6:47 AM, "Kessler CTR Mark J" <ma...@usmc.mil> wrote:
>Not sure how we changed a bitmap for variableRowHeight with the 3 below
>commits. I can look at the bitmaps when I get home. If it looks
>correct, how do I update the bitmap it uses to check with? Just copy
>it's compared one and push it back to the sdk?
Recommended practice is to run the test with -createImages option (and
then delete the associated png.xml and run it again without to verify that
there aren't any differences caused by that option).
Sometimes there are differences and the test needs upgrading to eliminate
those differences. If you just push the "compared one" then someone else
will be confused later if another change is required.
Did this run use a different player? If so, I wouldn't push a new bitmap,
I would investigate the differences, and if they are just anti-aliasing
differences, then I would add the png.xml file generated from the 11.1
player if that gets it to pass.
>
>-Mark
>
>
>-----Original Message-----
>From: flex.mustella@gmail.com [mailto:flex.mustella@gmail.com]
>Sent: Thursday, September 05, 2013 9:30 AM
>To: commits@flex.apache.org; KesslerConsulting@gmail.com;
>jmclean@apache.org; bigosmallm@gmail.com; fthomas@apache.org;
>cyrill.zadra@gmail.com
>Subject: Build failed in Jenkins: flex-sdk_mustella #290
>
>See <http://localhost:8080/job/flex-sdk_mustella/290/changes>
>
>Changes:
>
>[KesslerConsulting] FLEX-33710: Added index bounds checks.
>
>[KesslerConsulting] FLEX-33710: Changed the Range out of bounds check to
>a try/catch for a range error.
>
>[jmclean] 4.9 AIR support/11.9 Flash Player support
>
>
>
> [java] =====================================================
> [java] Failed:
> [java] =====================================================
> [java]
>gumbo/components/DataGrid/Properties/DataGrid_Properties_editable
>Editable_variableRowHeight Failed CompareBitmap(body:step 12) compare
>returned[object BitmapData]
> [java]
> [java]
> [java] =====================================================
> [java] Passes: 4
> [java] Fails: 1
> [java] =====================================================
> [java]
> [java]
> [java] Wrote summary to results.txt
> [java] Wrote failures to failures.txt
>
>BUILD FAILED
>C:\jenkins_slave\workspace\flex-sdk_mustella\mustella\build.xml:1538: The
>following error occurred while executing this line:
>C:\jenkins_slave\workspace\flex-sdk_mustella\mustella\build.xml:1656:
>Java returned: 1
>
>Total time: 1 minute 40 seconds
>
>ute shell' marked build as failure
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Alex Harui <ah...@adobe.com>.
On 9/6/13 4:29 PM, "Mark Kessler" <ke...@gmail.com> wrote:
>Well I found a better way to check it out if we want to find out when it
>first started. Look through the log on the image file. You put the
>image
>in on 9/5/1012 and it shows correctly... It shows the image without a
>line. Carol put in a new version on 12/21/2012 when she added that huge
>datagrid update and the line first appears there.
>
>-Mark
Cool. Thanks for digging that deep.
-Alex
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Mark Kessler <ke...@gmail.com>.
Well I found a better way to check it out if we want to find out when it
first started. Look through the log on the image file. You put the image
in on 9/5/1012 and it shows correctly... It shows the image without a
line. Carol put in a new version on 12/21/2012 when she added that huge
datagrid update and the line first appears there.
-Mark
On Fri, Sep 6, 2013 at 7:24 PM, Alex Harui <ah...@adobe.com> wrote:
>
>
> On 9/6/13 3:58 PM, "Mark Kessler" <ke...@gmail.com> wrote:
>
> >
> >My question would be do the tests no show the image compare failures or
> >further errors if a previous testID failed higher up in the same file.
> >Such as [4] which I fixed.
> Theoretically, the tests continue on unless an exception is thrown, so if
> the test in [4] was throwing, then yes it would hide this difference. If
> you really wanted to dig, you could grep the run log to see if that test
> ran or not.
>
> I did more digging and it looks like the other baseline in the test does
> have a black border from the bottom of the DG so I'm satisfied you've done
> everything correctly.
>
> Thanks for sticking with it.
>
> -Alex
>
>
>
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Alex Harui <ah...@adobe.com>.
On 9/6/13 3:58 PM, "Mark Kessler" <ke...@gmail.com> wrote:
>
>My question would be do the tests no show the image compare failures or
>further errors if a previous testID failed higher up in the same file.
>Such as [4] which I fixed.
Theoretically, the tests continue on unless an exception is thrown, so if
the test in [4] was throwing, then yes it would hide this difference. If
you really wanted to dig, you could grep the run log to see if that test
ran or not.
I did more digging and it looks like the other baseline in the test does
have a black border from the bottom of the DG so I'm satisfied you've done
everything correctly.
Thanks for sticking with it.
-Alex
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Mark Kessler <ke...@gmail.com>.
The line was in the image for months at least. Not sure why it wasn't
found until recently. But [1] and [2] were specifically about the DataGrid
itemEditor stopping in weird places while editing after having scrolled.
Which to mean would have been caused a line to appear since it would have
displayed incorrectly. If you run the test manually you can see it
displaying properly and before those other two [1][2] issues were fixed
would have been probably displayed partially through the bottom of the
datagrid border line.
The test itself says "verify variableRowHeight=true, different
columnWidth,editor should show in different size", it doesn't show it's
testing it to stop inbetween the rows.
My question would be do the tests no show the image compare failures or
further errors if a previous testID failed higher up in the same file.
Such as [4] which I fixed.
[1] https://issues.apache.org/jira/browse/FLEX-33665
[2] https://issues.apache.org/jira/browse/FLEX-33666
[3] mustella/tests/gumbo/components/DataGrid/SWFs/DataGrid_itemEditor.swf
[4] Editable_startItemEditorSesision_inValidParams
-Mark
On Fri, Sep 6, 2013 at 5:00 PM, Alex Harui <ah...@adobe.com> wrote:
>
>
> On 9/6/13 1:48 AM, "Mark Kessler" <ke...@gmail.com> wrote:
>
> >Yeah I got the horizontal line myself. I bet that was originally a bug
> >from before we fixed the scrolling itemEdit issue. The correct image
> >should not show the line through the box.
> Hmm. That scrolling position issue wasn't one of the three changes that
> could cause this failure was it?
>
> The conservative approach would be to locally revert all three changes and
> see if the test runs successfully. I set a breakpoint on the test's
> CompareBitmap.as:execute to see what else was on-screen at the time. I
> don't see a dark line in the DG, so I'm not sure it is the row separators.
> My concern is that the dark line is the border of the editor and that
> border is now hidden, but I also can't imagine which of the three changes
> could affect that either, and I would expect other tests to fail as well
> unless it is an issue with the editor border only when scrolled.
>
> -Alex
>
>
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Alex Harui <ah...@adobe.com>.
On 9/6/13 1:48 AM, "Mark Kessler" <ke...@gmail.com> wrote:
>Yeah I got the horizontal line myself. I bet that was originally a bug
>from before we fixed the scrolling itemEdit issue. The correct image
>should not show the line through the box.
Hmm. That scrolling position issue wasn't one of the three changes that
could cause this failure was it?
The conservative approach would be to locally revert all three changes and
see if the test runs successfully. I set a breakpoint on the test's
CompareBitmap.as:execute to see what else was on-screen at the time. I
don't see a dark line in the DG, so I'm not sure it is the row separators.
My concern is that the dark line is the border of the editor and that
border is now hidden, but I also can't imagine which of the three changes
could affect that either, and I would expect other tests to fail as well
unless it is an issue with the editor border only when scrolled.
-Alex
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Mark Kessler <ke...@gmail.com>.
Yeah I got the horizontal line myself. I bet that was originally a bug
from before we fixed the scrolling itemEdit issue. The correct image
should not show the line through the box.
-Mark
On Thu, Sep 5, 2013 at 10:58 PM, Alex Harui <ah...@adobe.com> wrote:
> Ran the tests myself. The bitmap shows a horizontal line through the
> reference image that is not in the result image. Please verify that you
> are seeing the same. You'll need to investigate whether that line should
> be there or not and if it is, figure out which code change resulted in the
> line not being there, or figure out why the test is now producing
> different results due to a flaw in the test.
>
> -Alex
>
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Alex Harui <ah...@adobe.com>.
Ran the tests myself. The bitmap shows a horizontal line through the
reference image that is not in the result image. Please verify that you
are seeing the same. You'll need to investigate whether that line should
be there or not and if it is, figure out which code change resulted in the
line not being there, or figure out why the test is now producing
different results due to a flaw in the test.
-Alex
On 9/5/13 7:49 PM, "Alex Harui" <ah...@adobe.com> wrote:
>
>
>On 9/5/13 3:43 PM, "Mark Kessler" <ke...@gmail.com> wrote:
>
>>Ok so I ran [1] with createdImages and the test passed. Deleted the two
>>XML files it created and ran [2] without the create option and it still
>>passed. Image looks fine. So do I upload the two new baseline images ?
>Well, what are the differences between the new baseline and the old one?
>You pretty much have to check the differences first. Most tests will
>consistently render and compare to a new baseline, but what we care about
>is whether that new baseline is exposing a bug. There could be a massive
>rowHeight change, for example. Use ImageDiffAIR to compare. Or send me
>the old and new baseline. I'll try to run that test now and look at the
>diffs if it fails. I forgot to do that earlier.
>
>-Alex
>
>>
>>[1] mini_run.sh -caseName=Editable_variableRowHeight
>>tests/gumbo/components/DataGrid/Properties/DataGrid_Properties_editable.m
>>x
>>ml
>>-createImages
>>
>>[2] mini_run.sh -caseName=Editable_variableRowHeight
>>tests/gumbo/components/DataGrid/Properties/DataGrid_Properties_editable.m
>>x
>>ml
>>
>>
>>-Mark
>>
>>On Thu, Sep 5, 2013 at 12:41 PM, Alex Harui <ah...@adobe.com> wrote:
>>
>>>
>>>
>>> On 9/5/13 6:47 AM, "Kessler CTR Mark J" <ma...@usmc.mil>
>>>wrote:
>>>
>>> >Not sure how we changed a bitmap for variableRowHeight with the 3
>>>below
>>> >commits. I can look at the bitmaps when I get home. If it looks
>>> >correct, how do I update the bitmap it uses to check with? Just copy
>>> >it's compared one and push it back to the sdk?
>>> Recommended practice is to run the test with -createImages option (and
>>> then delete the associated png.xml and run it again without to verify
>>>that
>>> there aren't any differences caused by that option).
>>>
>>> Sometimes there are differences and the test needs upgrading to
>>>eliminate
>>> those differences. If you just push the "compared one" then someone
>>>else
>>> will be confused later if another change is required.
>>>
>>> Did this run use a different player? If so, I wouldn't push a new
>>>bitmap,
>>> I would investigate the differences, and if they are just anti-aliasing
>>> differences, then I would add the png.xml file generated from the 11.1
>>> player if that gets it to pass.
>>>
>
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Alex Harui <ah...@adobe.com>.
On 9/5/13 3:43 PM, "Mark Kessler" <ke...@gmail.com> wrote:
>Ok so I ran [1] with createdImages and the test passed. Deleted the two
>XML files it created and ran [2] without the create option and it still
>passed. Image looks fine. So do I upload the two new baseline images ?
Well, what are the differences between the new baseline and the old one?
You pretty much have to check the differences first. Most tests will
consistently render and compare to a new baseline, but what we care about
is whether that new baseline is exposing a bug. There could be a massive
rowHeight change, for example. Use ImageDiffAIR to compare. Or send me
the old and new baseline. I'll try to run that test now and look at the
diffs if it fails. I forgot to do that earlier.
-Alex
>
>[1] mini_run.sh -caseName=Editable_variableRowHeight
>tests/gumbo/components/DataGrid/Properties/DataGrid_Properties_editable.mx
>ml
>-createImages
>
>[2] mini_run.sh -caseName=Editable_variableRowHeight
>tests/gumbo/components/DataGrid/Properties/DataGrid_Properties_editable.mx
>ml
>
>
>-Mark
>
>On Thu, Sep 5, 2013 at 12:41 PM, Alex Harui <ah...@adobe.com> wrote:
>
>>
>>
>> On 9/5/13 6:47 AM, "Kessler CTR Mark J" <ma...@usmc.mil>
>>wrote:
>>
>> >Not sure how we changed a bitmap for variableRowHeight with the 3 below
>> >commits. I can look at the bitmaps when I get home. If it looks
>> >correct, how do I update the bitmap it uses to check with? Just copy
>> >it's compared one and push it back to the sdk?
>> Recommended practice is to run the test with -createImages option (and
>> then delete the associated png.xml and run it again without to verify
>>that
>> there aren't any differences caused by that option).
>>
>> Sometimes there are differences and the test needs upgrading to
>>eliminate
>> those differences. If you just push the "compared one" then someone
>>else
>> will be confused later if another change is required.
>>
>> Did this run use a different player? If so, I wouldn't push a new
>>bitmap,
>> I would investigate the differences, and if they are just anti-aliasing
>> differences, then I would add the png.xml file generated from the 11.1
>> player if that gets it to pass.
>>
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Mark Kessler <ke...@gmail.com>.
Ok so I ran [1] with createdImages and the test passed. Deleted the two
XML files it created and ran [2] without the create option and it still
passed. Image looks fine. So do I upload the two new baseline images ?
[1] mini_run.sh -caseName=Editable_variableRowHeight
tests/gumbo/components/DataGrid/Properties/DataGrid_Properties_editable.mxml
-createImages
[2] mini_run.sh -caseName=Editable_variableRowHeight
tests/gumbo/components/DataGrid/Properties/DataGrid_Properties_editable.mxml
-Mark
On Thu, Sep 5, 2013 at 12:41 PM, Alex Harui <ah...@adobe.com> wrote:
>
>
> On 9/5/13 6:47 AM, "Kessler CTR Mark J" <ma...@usmc.mil> wrote:
>
> >Not sure how we changed a bitmap for variableRowHeight with the 3 below
> >commits. I can look at the bitmaps when I get home. If it looks
> >correct, how do I update the bitmap it uses to check with? Just copy
> >it's compared one and push it back to the sdk?
> Recommended practice is to run the test with -createImages option (and
> then delete the associated png.xml and run it again without to verify that
> there aren't any differences caused by that option).
>
> Sometimes there are differences and the test needs upgrading to eliminate
> those differences. If you just push the "compared one" then someone else
> will be confused later if another change is required.
>
> Did this run use a different player? If so, I wouldn't push a new bitmap,
> I would investigate the differences, and if they are just anti-aliasing
> differences, then I would add the png.xml file generated from the 11.1
> player if that gets it to pass.
>
Re: Build failed in Jenkins: flex-sdk_mustella #290
Posted by Alex Harui <ah...@adobe.com>.
On 9/5/13 6:47 AM, "Kessler CTR Mark J" <ma...@usmc.mil> wrote:
>Not sure how we changed a bitmap for variableRowHeight with the 3 below
>commits. I can look at the bitmaps when I get home. If it looks
>correct, how do I update the bitmap it uses to check with? Just copy
>it's compared one and push it back to the sdk?
Recommended practice is to run the test with -createImages option (and
then delete the associated png.xml and run it again without to verify that
there aren't any differences caused by that option).
Sometimes there are differences and the test needs upgrading to eliminate
those differences. If you just push the "compared one" then someone else
will be confused later if another change is required.
Did this run use a different player? If so, I wouldn't push a new bitmap,
I would investigate the differences, and if they are just anti-aliasing
differences, then I would add the png.xml file generated from the 11.1
player if that gets it to pass.
>
>-Mark
>
>
>-----Original Message-----
>From: flex.mustella@gmail.com [mailto:flex.mustella@gmail.com]
>Sent: Thursday, September 05, 2013 9:30 AM
>To: commits@flex.apache.org; KesslerConsulting@gmail.com;
>jmclean@apache.org; bigosmallm@gmail.com; fthomas@apache.org;
>cyrill.zadra@gmail.com
>Subject: Build failed in Jenkins: flex-sdk_mustella #290
>
>See <http://localhost:8080/job/flex-sdk_mustella/290/changes>
>
>Changes:
>
>[KesslerConsulting] FLEX-33710: Added index bounds checks.
>
>[KesslerConsulting] FLEX-33710: Changed the Range out of bounds check to
>a try/catch for a range error.
>
>[jmclean] 4.9 AIR support/11.9 Flash Player support
>
>
>
> [java] =====================================================
> [java] Failed:
> [java] =====================================================
> [java]
>gumbo/components/DataGrid/Properties/DataGrid_Properties_editable
>Editable_variableRowHeight Failed CompareBitmap(body:step 12) compare
>returned[object BitmapData]
> [java]
> [java]
> [java] =====================================================
> [java] Passes: 4
> [java] Fails: 1
> [java] =====================================================
> [java]
> [java]
> [java] Wrote summary to results.txt
> [java] Wrote failures to failures.txt
>
>BUILD FAILED
>C:\jenkins_slave\workspace\flex-sdk_mustella\mustella\build.xml:1538: The
>following error occurred while executing this line:
>C:\jenkins_slave\workspace\flex-sdk_mustella\mustella\build.xml:1656:
>Java returned: 1
>
>Total time: 1 minute 40 seconds
>
>ute shell' marked build as failure