Oxygen's new window-move feature fatal for KJumpingCube game

[Ian Wadham]

On Thursday 17 June 2010 10:34:31 am Hugo Pereira Da Costa wrote:
> Hello Ian,
> 
Hello Hugo,

Thank you very much for your excellent explanation
of what is going on.  I understand where you are coming from,
having once been a library and OS designer (but that was back
in C and assembly language days).  However, as an application
writer, I am left feeling rather uneasy about some changes that
happen in libraries.

Before I continue, I should say that I admire the concern,
involvement and responsibility you personally are showing.
Unfortunately that is not always the way things are.

Your explanation of the problems involved in implementing the
new window-move feature in a general way is extremely interesting,
but it begs some important questions.  Why is the feature necessary
and is its benefit worth the possible disruption to existing applications
and the extra work for their maintainers and end users?

Such questions apply IMHO to many new core features, not just this one.

I know, "coder decides" and all that, but I cannot escape the impression
sometimes that such decisions are made on the basis of "cool", neat
visual effect, etc. and the benefits for a few "advanced" users.

I wonder how much the impact on others, such as library users
(i.e. application developers) and end-users is considered.

On this occasion, Parker and Hugo rushed in and fixed the problems
for all KDE Games, promptly and without fuss.  Their professionalism
is to be applauded.

Nevertheless, the bug report on KJumpingCube caused Wyatt
(the reporter) and me some lost hours emailing, googling and
reading blogs, even though it turned out there was no problem
to fix.  I would have rather spent that time on using my computer
to prepare some Science talks I am giving or even playing with
my grandchildren ... :-)

I would not be making a fuss, except that this is not the first time
I have been in this situation.  On the last KDE release, for instance,
an implementation change in KConfigGroup proved fatal for the
Kolf game (and it took a while to diagnose that).  The game was
threatened with extinction (removal to "unmaintained"), having no
maintainer at the time, but there was no help with diagnosing the
problem offered from this group, despite appeals.

In the end I nailed it (I'm an old troubleshooter from way back),
but now (somewhat unwillingly) I am the "maintainer" of Kolf ... :-(

I know ... games are not important in the general scheme of things.
Nevertheless they all-too-frequently point up problems in libraries.
They are good test-beds for some features.

> No real blame to the applications: they were there first, and Qt does not
> require (in general) that you eat an event when you do something with it
> (although it does, sometimes).
> 
Thank you, Hugo.  I really appreciate your concern for us library users.

> Note that parker's commit to kjumpingcube just addresses that (so that the
>  bug is actually fixed twice)
> 
In a way, perhaps, KJumpingCube was getting its just treatment this time
around.  I did not write it, but took over maintenance to save it going
into the bit-bucket when KDE 4 and Qt 4 came along.

Back in KDE 2 days, KJCube had a bug that could bring down the KDE
desktop and X.  If you selected largest board-size and high difficulty level,
it would go away and "think" for a very long time and, slowly but surely,
all widgets and windows would slow down and stop responding and the
mouse pointer would eventually freeze up.  It was a memory leak of course.

So now perhaps KDE ihas got its revenge on KJCube ... :-)

All the best, Ian W.
KDE's oldest developer, but not an "advanced" user ... :-)