On Jan 1, 2015 9:28 PM, "Alex Harui" <aha...@adobe.com> wrote: > > I haven’t looked at the diffs, but usually I want to understand why > something is giving sub-pixel diffs before cutting new baselines. Once it > was because we changed the parenting of child objects in the skin. That > made the player choose slightly different color values. Once it was > because we added or removed rounding of pixel values to integer values. > That made some things look sharper because single pixel lines were on > pixel boundaries. But just as often, there is some subtle bug that breaks > some customer. Some folks have eagle-eyes for slight differences in > rendering. > > Note also that I think the failures on the main Mustella run are due to > player issues. Maybe if Erik reboots, a lot of these issues will go away. >
You are right. I looked at the diffs using the ImageDiff app. There is an artifact one pixel hight and a few pixels wide in most of the ActionBar bitmaps. I have not been able to figure out where this is coming from. I did not touch this component. Thanks, Om > -Alex > > On 12/31/14, 1:26 PM, "OmPrakash Muppirala" <bigosma...@gmail.com> wrote: > > >The other set of failing Mustella tests are for ActionBar. Almost all of > >the failures (except for 4) are sub-pixel failures. Should I just > >recreate > >the baselines for these ones? Not sure how else to fix this. > > > >Thanks, > >Om > > > >On Wed, Dec 31, 2014 at 12:52 PM, OmPrakash Muppirala > ><bigosma...@gmail.com> > >wrote: > > > >> Here is some information on the failing tests: > >> > >> BusyIndicator: > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_weak_ref_test Failed AssertPropertyValue(body:step 1) > >>bi02.isInMemory > >> true != false > >> > >> The BusyIndicator used to be a component with the visuals inside itself. > >> I split it up into a component and its skin. > >> It appears that components with spark skins do not get removed from > >>memory > >> that easily. I have try setting mx_internal::skinDestructionPolicy = > >> "auto" that does not seem to do the trick. > >> > >> Does all spark skinned components have this problem? > >> > >> > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_symbolColor Failed CompareBitmap(body:step 2) compare > >>returned[object > >> BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_symbolColor_default Failed CompareBitmap(body:step 2) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_SDK_29330 Failed CompareBitmap(body:step 1) compare returned[object > >> BitmapData] > >> > >> The BusyIndicator, while shown must rotate. When not shown, i.e. when > >> visible=false, will be in a non-rotating state. As the toggle from > >>hidden > >> to visible happens, something with new skin causes the component to be > >>not > >> shown. The component itself works fine, but for some reason, these > >>tests > >> are breaking. > >> > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_default Failed CompareBitmap(body:step 1) compare > >>returned[object > >> BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_dif_width_height Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_dif_height_width Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_comp_size_0x0 Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_comp_size_25x25 Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_comp_size_100x100 Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_comp_size_200x200 Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> [java] > >> > >>mobile/components/BusyIndicator/integration/BusyIndicator_Integration_tes > >>ter > >> BI_size_comp_scaleXandscaleY Failed CompareBitmap(body:step 1) compare > >> returned[object BitmapData] > >> > >> These are tests related to sizing. Again, the component in normal usage > >> does get sized correctly, but these tests are failing. Not sure if the > >> tests need modification. > >> > >> > >> On Wed, Dec 31, 2014 at 3:29 AM, Harbs <harbs.li...@gmail.com> wrote: > >> > >>> If you can give me some pointers, I’d be happy to take a look. > >>> > >>> Harbs > >>> > >>> On Dec 31, 2014, at 3:47 AM, OmPrakash Muppirala <bigosma...@gmail.com > > >>> wrote: > >>> > >>> > I have spent more time than I wanted to on the failing Mustella > >>>tests. > >>> I > >>> > have managed to fix quite a few, but there are a still a bunch of > >>>cases > >>> > that are failing. They don't look that hard to fix, but for the > >>>life of > >>> > me, I have not been able to figure out what's happening. > >>> > > >>> > Can someone take over for me? I am going on vacation for a couple of > >>> weeks > >>> > and it is going to be hard for me to work on this when I am gone. > >>> > > >>> > Alternatively, we can disable these tests (gasp!) for the time being. > >>> > > >>> > Ideas, volunteers welcome. > >>> > > >>> > Thanks, > >>> > Om > >>> > >>> > >> >
