Please use -createImages to generate baselines. There can be faulty tests where createImages produces a different image than at compare time. We want to use a consistent technique so we know how things get generated.
It should be rare for a baseline to be missing. I believe you've said you've been deleting them so there's a chance you are causing a problem by doing that. -createImages always overwrites current images and SVN won't check them in unless it sees an actual difference. On 8/23/12 1:29 PM, "Carlos Rovira" <carlos.rov...@codeoscopic.com> wrote: > Hi Alex, > > for this kind of message: "Failed CompareBitmap(body:step 3) Baseline > image could not be read. Created image file as a .bad.png." > > I'm renaming the created image .bad.png to the expected image to pass. The > problem here is that there's no image, so I think the one created could be > valid. > > If some of the things I post here is not right let me know... > > 2012/8/23 Alex Harui <aha...@adobe.com> > >> >> >> >> On 8/23/12 9:29 AM, "Carlos Rovira" <carlos.rov...@codeoscopic.com> wrote: >> >>> Hi Alex, >>> >>> >>> 2012/8/23 Alex Harui <aha...@adobe.com> >>> >>>> >>>> I would use the ImageDiffAir tool to figure out what is different, then >>>> examine the test to determine which image is the right one. >>>> >>>> >>> The screen shot is from ImageDiffAir tool. >> Yes, you can use the getPixel button and/or the pixel reading checkbox to >> determine the pixel values at certain coordinates to determine in exact >> detail the differences. >>> >>> >>> >>>> If the red square asset got changed and is smaller then maybe the >> baseline >>>> wasn't refreshed after the asset changed. >>>> >>> >>> I tried more than one time and remove the images to be generated again, >> so >>> I think there's no much to do there >> I thought you'd already generated new baselines so it was making me think >> that the test was unstable. If that's not the case, then you're probably >> ok >> with just cutting a new one. >>> >>> >>>> >>>> It could also be differences in rendering between platforms and machines >>>> and >>>> then we'll have to decide to exclude it or find a way to fix it. >>>> >>>> >>> I think this is what is happening since from the capture I sent not only >>> red square is different, "Name" text report changes too. >> Yes, but the position of Name might be dependent on red square size. >>> >>> As I'm new to mustella and already not have too much experience I'd like >>> that someone point me how to proceed with this one. >>> >> If the new baseline seems stable (a second run shows no differences) then >> just check it in for now. If there is a platform/machine difference we'll >> find those as others run these tests and then try to figure out how to >> stablize the test. >> >> It gets trickier for tests of effects because timing can affect the >> rendering, but most tests don't have moving parts so just getting the >> folder >> to run with zero failures on your machine is good enough for now. >> >> -- >> Alex Harui >> Flex SDK Team >> Adobe Systems, Inc. >> http://blogs.adobe.com/aharui >> >> > -- Alex Harui Flex SDK Team Adobe Systems, Inc. http://blogs.adobe.com/aharui
