[Kde-games-devel] Regression testing

[Ian Wadham]

With all the talk there has been about Kolf, I was getting a bit
panicky about upgrading to Qt 4.5 and the latest KDE trunk and
finding that maybe my three games had regressed.  I have been
working off KDE 4.2 (approx) and Qt 4.4.0.  So I finally bit the
bullet, but before I did I hacked a bit at David Faure's cs and cb
functions (in .bashrc), which I use for development work.

Now I have built and installed Qt 4.4.0, Qt 4.5.1 (qt-copy),
KDE 4.2 trunk and KDE 4.3 trunk, with the latest KDE Games
trunk and I can mix and match library versions (e.g. kv 4.3; cb
to build with KDE 4.3 libraries).  kv and qv are my versioning
functions.

Anyway, is there anything anybody would like me to try with
this setup?

KGoldrunner, Kubrick and KJumpingCube were OK with
KDE 4.3 trunk and Qt 4.5.1 (phew!).  So were Killbots, KBounce
and Kollision.  Couldn't see any problems with KMines, but
KPat cut a corner out of a card once in about an hour's play.

Kolf was a mess as expected.  Not only did I get blank layouts
for holes, but the putter did not work properly and some holes
started with water in the middle of the screen and my ball in it!

I hope something is being done on kde-core-devel about the
Kolf situation.  Where I come from (professionally, before I
retired), the order of priority, from highest to lowest, was:

   1. Preserve functionality and reliability for end users.
   2. Preserve stability, functionality and reliability of supporting
       software for application programmers.
   3. Upgrade to the latest supporting software, but only after
       extensive regression testing.

These rules were based on the greatest good of the greatest
number ... N users > N application programmers > N system
programmers.

In the present case, we would NOT abandon Kolf.

Rather we would abandon Qt 4.5 for now (i.e. hold it over till
KDE 4.4).

All the best, Ian W.