The Bad News: =========== There are a few small errors in PyKDE-3.7-3 that ONLY affect KDE3.1.3, the first being you can't actually build PyKDE for KDE3.1.3 (it builds for KDE3.1.2 instead, which isn't really a problem - it still runs under KDE3.1.3). If you could build against KDE3.1.3, you'd find the sip files for dcop aren't included in the build (they're there - they just don't get built or imported by later modules). There's one other small versioning error, and I overlooked the fact that KDE includes kaccelmanager.h in both kdecore and kdeui. All fixed already.
If that's unclear, PyKDE-3.7-3 will build under KDE 3.1.3 and run, but it won't allow use of new KDE 3.1.3 features (I'm not sure there are any really) from PyKDE. The Good News: ============ PyKDE-3.7-3 builds against the latest sip/PyQt snapshots (20030801), so sip 3.8 compatibility shouldn't be hard to come by. This is especially good news since there appears to be a sip problem in the kfile module, and Phil will likely patch the snapshot instead of going back to rework sip 3.7 (sip 3.8 isn't too far off). Both sets of problems above are due to inadequete testing. The "bad news" stuff was due to an oversight in updating the version list in build.py, and the kfile problems weren't caught because the testing for that module only covers handwritten code at present and assumes the machine generated code (which is almost everything) is correct. 95% of the time that's a good assumption. I've tended to put testing where it makes the most difference first, and that's economically sound, but doesn't always guarantee correct software. Rant: ==== I've done a few responses to people asking about how they can contribute to PyKDE, and they all mention testing as a priority. It takes me about 2 to 3 minutes to generate all of the sip files completely versioned and (almost) ready to build for a new release (and, no, I'm not kidding at all about the 2 to 3 minutes - actually it takes another 10 -15 minutes to update the project file too). It takes considerably longer to implement good testing and I'd really appreciate help in that area. I got a copy of "Refactoring" by Fowler in April, and from that, the general philosophy of test-driven development, and 8 years as a test engineer and manager working with automated hardware test systems, I'm getting serious to fanatical about having testing in place for *everything*. You should be too. I'm real tired of software bugs, even if they're my own. Even if the problem ends up being in sip this time, it's still my responsibility to test sip-produced code and verify sip against PyKDE (and let Phil know when problems occur) - PyKDE is much harder in terms of what it demands from sip than any other bindings. That's one reason that, even before tonight's new bug report, I've set up a box for testing and building snapshots, and when appropriate I'll be posting PyKDE snapshots for testing to SourceForge. In Closing: ======== There will be another PyKDE release as soon as fixes are in place for the bugs mentioned above. It may also include some additional features, as I have most of the operator overload stuff added to PyKDE (and yes, I've done at least some testing on the new features). Jim _______________________________________________ PyKDE mailing list [EMAIL PROTECTED] http://mats.imk.fraunhofer.de/mailman/listinfo/pykde
