release readiness

[Boudewijn Rempt]

We've just released the 4th beta, so it's time to check the current state of Krita. We've made some good progress, but I'm not entirely elated, to be honest... We're far from an excellent release yet.

=== Unit tests ===

From zero failing unittests, we're back at five!

         80 - krita-image-KisCloneLayerTest (Failed)
        103 - krita-image-KisWalkersTest (Failed)
        104 - krita-image-KisAsyncMergerTest (Failed)
        144 - krita-image-KisActionRecorderTest (Failed)
        145 - krita-image-KisProcessingsTest (Failed)

As long as these tests fail, we simply cannot release!

=== Critical bugs and crashes ===

We have 3 critical bugs left:

285441 srikanth.tulasiram at gmail.com Downloading presets from GHNS does no longer work
262324 lukast.dev at gmail.com krita produces artefacts when upscaling
217292 boud at valdyas.org Ability to paint transparency masks 

2 grave bugs, 4 major bugs, 31 crashes and a grand total of 111 bugs. Functionality that is currently completely broken or seems unusably broken to the user is:

* GHNS integration
* Preset creation (previews are broken, saving presets with a name of a preset that was deleted earlier on)
* Working with masks

I also found memcheck errors in the jpg filter that worry me a lot: https://bugs.kde.org/show_bug.cgi?id=287401

=== Memory leaks ===

Tobias Hagberg reported a memory leak bug. He worked with 2.3, but it turns that wether his leak is the same as our leak, we do have a leak in git master. Check the attached PNG image. It shows the memory consumption of Krita over time. What I did was load an image, convert it to cmyk, save it as tiff, close the image (but not krita), rinse and repeat. The dark red bottom band is tile data. You can see that we do release memory when closing an image, but not nearly all of it! And in fact, when we close the image for the second time, some memory is released, but again, not all, until the application quits. 

See bug https://bugs.kde.org/show_bug.cgi?id=287344 for massif logs, the output of our own memory checker and more details.

I am strongly minded to classify this bug as critical. It ties in with some of the crashes Silvio Grosso reported, since those are also about memory leaks and I am certain that this is a regression.

-- 
Boudewijn Rempt
http://www.valdyas.org, http://www.krita.org, http://www.boudewijnrempt.nl
-------------- next part --------------
A non-text attachment was scrubbed...
Name: massif.png
Type: image/png
Size: 575620 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kimageshop/attachments/20111124/a9c8b626/attachment-0001.png>