The first thing in the previous sprint was to make the new PySide release (0.3.2). This release is the first one to include the new-style signal and slot architecture, support for Qt 4.7 beta (including support for QML/Qt Quick). Of course, a hefty amount of bugs were fixed as well.

After the release, the most of the sprint was spent in implementing the API 2.0. This work got ALMOST finished, with the keyword argument implementation just barely slipping to the next sprint.

So, in the upcoming sprint, the first thing is to finish the API 2 work – the team will send a separate notification when that’s done. We discussed about making a new release but since the API 2 keyword arguments are still missing and we have plenty of new bug reports, we opted to let the dust settle a bit and fix some of the most outstanding bugs. In addition to that, we’ll also begin working on the Qt Mobility APIs, which are especially relevant on the Maemo/MeeGo platforms.

We’ll probably make the next release in the beginning of the next sprint, i.e. in 2-3 weeks from now. That release would then be the last one to have backwards-incompatible changes before the long-awaited 1.0 launch! Once that version is out, the focus will be on bugfixing, although work should resume on the Python 3 front as well.

Matti Airas

Now developers can use signals and slots in a pythonic fashion. More info can be found at documentation page. Another point covered by this release is the ability to work with or without the so called “protected hack”. When enabled it reduces library size but doesn’t allow PySide to be compiled under Windows environment. This time it is possible to turn it on or off according to the needs.

- New-style Signal/Slot;
- Lots of bugs fixed;
- Generator works without protected hack;

Enjoy it!

In the past sprint, the main new content was support for Qt 4.7 beta 1. This was successfully done: if you want to try out some Declarative UI goodness with a Python backend, get the latest and greatest Qt and PySide and enjoy!

In other news, a horde of bugs was squashed, and last but not least, the biggest blocker for porting PySide to Windows (the “#define protected public” hack) is not required any longer. According to our understanding, no major issues preventing a complete Windows port should exist any longer. May the hacking commence!

In the sprint starting today, the very first thing is to make a new release to get the latest features in the hand of developers and platform porters not interested in working with the git snapshots. The release should happen on Monday, at the latest.

The major new functionality, actually filling the rest of the sprint, is the implementation of API 2. We hope to finish this completely within this sprint (two weeks), after which we should have all Python 2.x functionality present for a 1.0 release.

Since the sprint was completely filled by API 2 stuff, Python 3 work has been postponed after that. We hope to start investigating that after this sprint.

Matti Airas

Here’s again a brief update on what we’ve been doing, what’s going on at the moment, and what’s been planned for the near future.

In our previous sprint, which ended last Wednesday, the big new thing was support for new-style signals and slots. That’s already available in Gitorious, although the class name capitalization might still need to be updated according to re-opened PSEP 100. Also, a huge bunch of bugs were fixed.

In the present sprint, we’ve already fixed 9 bugs and work to implement the new features of Qt 4.7 is ongoing. We expected to have time only to barely touch the Qt 4.7 stuff but the team is absolutely on fire and there’s actually a fighting chance that we have full Qt 4.7 support done by the end of this sprint! Assuming this will happen, there’s probably going to be a new release next week (although I didn’t discuss this with the team yet).

We’ll be continuing the intense bugfixing mode in the near future but also work on new features. Next things to work on will be investigating Python 3 and API 2 (PSEP 101) implementation and then actually implementing them. My personal wish is to freeze the API at that point and begin a maturization period to make PySide really production-quality. As discussed, this wouldn’t mean the API development stops there but that we would just need to find ways to implement any future changes without breaking existing software.

On the bindings front, Qt Mobility API is the next thing to tackle once Qt 4.7 is there. This development will probably happen side-by-side with the API implementation work.

Matti Airas

What’s changed since 0.3.0?

With the technical preview release a few weeks ago we moved from a Boost based source code to a CPython one. This move was necessary to reduce our footprint. Now we have the first official release of this new PySide based on Shiboken (CPython) generator. Since the technical preview the main points are:

- Bindings for the last missing modules: Phonon, QtScript, QtScriptTools, QtMaemo5, QtHelp, QtMultimedia, QUiTools and QXmlPatterns;
- Fixed some memory and reference leaks;
- Lots of bugs fixed;
- Revamped documentation.

For more Qt 4.6 information: http://qt.nokia.com/doc/4.6/qt4-6-intro.html

We hope you enjoy this new release! Feel free to contribute with comments,
bug reports and suggestions.

Regards,
PySide Dev Team

The release will also imply updated build instructions, reference docs, and website. These have been lagging a bit behind but the situation will finally be remedied.

Right after the release, we intend to prepare more comprehensive performance comparisons and graphs along with some functional benchmarking. In the upcoming sprints, the immediate items to work on would be Qt 4.7 support, new-style signals and slots as well as begin the API 2 and Python 3 work. The exact ordering might be different, however, and I wouldn’t want to make any schedule commitments either.

Matti Airas

As PySide-Shiboken is getting ready for a release we have produced some graphs that show startup time and disk size usage. They are good numbers since we did not spend time doing optimizations on PySide-Shiboken source code yet.

Load time (QtCore + QtGui) takes less than a second on a N900. In a desktop the difference is not noticeable.

Startup time graph

The module size has dropped drastically from Boost implementation to this new one based on CPython. Now we are closer to PyQt4. As you know this size reduction also affects loading time.

Size comparision graph.

During the next weeks we will be working on bugfixing stuff and new modules. After that it will be possible to focus on performance and size reduction.

After a long wait, we have finally gotten the Shiboken generated PySide bindings in a good enough shape to do a proper release with tarballs and all the required niceties, instead of rough git urls. Keep in mind that this is an alpha release, or a “technical preview” as the kids in my lawn are used to say, some modules are missing and bugs are not hard to find.

This PySide release contains bindings for the following modules:

• QtCore
• QtGui
• QtNetwork
• QtWebKit
• QtSvg
• QtXml
• QtTest
• QtOpenGL
• QtSql

Feel free to try your Python code with this version of PySide. You know how it works: more users == less bugs. As long as you guys tell us about the  bugs: http://bugs.openbossa.org

A noteworthy fact is the first release of the Shiboken C++ binding generator.  Yay! We strongly encourage it’s use to produce also non-Qt C++  bindings for Python, it will make your life easier. If it turns out to be not so  easy, come and talk with us on #pyside channel (Freenode).

Here follow everything needed for this release, in order of compilation.

API Extractor, version 0.4.0
http://www.pyside.org/files/apiextractor-0.4.0.tar.bz2

Generator Runner, version 0.4.0
http://www.pyside.org/files/generatorrunner-0.4.0.tar.bz2

Shiboken, version 0.2.0
http://www.pyside.org/files/shiboken-0.2.0.tar.bz2

PySide, version 0.3.0
http://www.pyside.org/files/pyside-qt4.6+0.3.0.tar.bz2

You will notice that the size of the generated bindings for the Qt modules in no way resemble the figures from the Boost.Python version. And talking about Boost: packagers, be aware that no component depends on Boost::Graph anymore. Yay²!

P.S.: cgoncalves, thanks for all the code and for the reminding us to put the announcement here as well. And to everyone else that provided bug reports, patches, beers, etc.