[KDE/Mac] Giving up on kdeinit4 and friends

["René J.V. Bertin"]

On Aug 29, 2014, at 11:32, Mario Fux wrote:

>> OS X into some sort of working shape that is independent of KDE
>> core processes and non-portable objects like KCookieJar.

What's non-portable in kcookiejar? It seems to work just fine for me - provided kded4 is running. Maybe it ought to receive a Mac backend just like kwallet?

> I still think it's important to see that first kdeinit(4) is an open question 
> even for the kdelibs/KF5 people. And second feature development for kdelibs4 
> ended years ago and eventhough there is still bugfixing done for kdelibs4 most 
> of the power goes into KDE Frameworks 5 as the successor of kdelibs4. And 
> let's not forget that there is not a huge amount of manpower of KF5 as well...

What everyone also needs to realise is that with the current state of things, KDE development on Mac has to take place in KDE 4 for the vast majority. Also, we don't have the (man)power of distribution groups ([K]Ubuntu, Arch, OpenSUSE, ...) behind us who can provide playgrounds like the Neon5 project.

At the same time, the organisation/implementation of these mailing lists don't help integration, and rather favour the them vs. us spirit that Mario clearly despises so much. At the same time, I for one am not looking towards joining yet another mailing list just in order to get feedback (e.g. on things I cross-posted) from people on kde-devel - and the forums aren't that ... reactive.
I haven't tried IRC yet, though - have you, Ian? (In fact, I did for some Calligra/Krita things I came across: worked quite well.)

> of KDE in the same way as everybody else. But we can't expect from people that 
> work hard on KF5 on Linux and don't know anything about Mac OS X to help us. 

Not for things that are really specific to OS X, but we have our own resources for that. What we do need help with is the KDE internals, knowing how they're supposed to work, and how we might be able to work around incompatibilities with the way things are done on OS X. Remember, we're not reimplementing KDE from scratch. We're trying to get the existing code to work.
As an example, I recently came across crashes in kmail that I finally traced back to a call to KMenu::addTitle on a menu belonging to the Mac style toplevel application menubar. The crash happens in Qt code, and I think the bug is actually at that level. I came up with a workaround at the application level (I asked suggestions for that on the kde dev. forum but the post didn't even get read) but I'd have loved to modify KMenu::addTitle to do things differently when targeting items in the menubar. I just don't know how to detect that - but a "core KDE dev" should have been able to tell me if it's possible, and how. And that wouldn't even have taken a lot of time, because I'd still be doing the brunt of the work myself.
>> 
>> developer: not full time help, just a few pointers and some reviews of
>> finished work.  Eventually, even though he must have been up to his

As I said ...

> Too bad. But I'm sure this guy had his reasons.

Up to over his eyeballs in the same meeting? :)

> I hope you find back your joy in hacking on KDE code and send you some Swiss 
> hugs
If you don't mind, there are other things from Switzerland that I prefer :P

R.