[Kde-games-devel] KMahjongg frameworks branch

[Ian Wadham]

Hi Frederik,

On 16/11/2015, at 6:51 PM, Frederik Schwarzer wrote:
> Am Montag, 16. November 2015, 17:46:28 schrieb Ian Wadham:
>> On 07/11/2015, at 10:58 AM, Frederik Schwarzer wrote:
> 
> @Albert: comments like these. :D
> 
>> There is something I should tell you.  Back a few years ago, there
>> was a major branch in KMahjongg that never got merged into master. 
> 
> And that's exactly why I stopped working on KShisen's internals as I 
> had initially planned a few months ago and started to merge all 
> frameworks branches. If branches are open for too long, things get 
> complicated. Or to quote the Hagakure: "It is bad when one thing 
> becomes two." *coughs*

Hmmm… I had not heard of the Hagakure before…  The author may
have been referring to Japan having two rulers at the time…  Our case
may not be so serious.  The bugs are small ones and could perhaps
be ignored on a first run through…  They have been around for years
anyway…  The switch to QGraphicsView is a major development
landmark and should not be missed.

> So, first of all, at least I do not think, you are at fault here.

Well, I do feel a bit guilty, mainly because I said I would do something and
then did not finish it.  At the time it seemed that the many problems with Qt4,
KDE 4, Qt5 and Frameworks on the Apple OS X platform had higher priority
(for me).  Those problems are still unsolved... and there is just one guy left
working on them... and he is not a core KDE developer.

> For most of us, this is a hobby. That does not mean, we can do harm here 
> and run away laughing ... but it means that life in the form of other 
> things can get into the way. That's just how it is.
> 
> But now we need a decision here. You say it would be better to merge 
> the qgraphics branch into the frameworks branch and then merge all at 
> once into master once it is finished?

The plan, proposed by Laurent Montel, who was previously working on
the ports of games to Frameworks (including KMahjongg) --- the plan was
to merge:
    a. master ===> qgraphic --- DONE, see my commits on qgraphic branch
    b. frameworks ===> qgraphic --- then fix known bugs
    c. qgraphic ===> frameworks --- finish porting to Frameworks
    d. frameworks ===> master --- release as a KF5 app with QGraphicsView

I have completed part of step 2 and I propose that we skip dealing with
the rest of Jan-Peter's fixes for known bugs for the moment (but we should
come back to them later).  So proceed to steps 3 and 4.

> Usually I would object against that in favour of two smaller merges 
> with less impact over the period of two releases. But since you say, 
> things worked quite fine there, maybe it's worth a try?

Note that (in the "Mahjongg archaeology" thread), Roney reported a
successful merge of qgraphic to frameworks, with no difficulties.  So
I think it is safe for you to try step c, Frederik.

The qgraphic branch has been well-tested (by me) as reported in the
"Reviving KMahjongg port to QGraphicsView" thread:
https://mail.kde.org/pipermail/kde-games-devel/2015-January/011521.html

> Would you be willing to help out after the merge or are you retired 
> for good from getting your hands dirty (I thought I read something 
> about that a year ago?)

I said back then that I would continue pottering around, but then I
became embroiled in the KDE-and-KF5-on-OS-X fracas.  That was
a very bad experience for me, mainly because of the unhelpful
attitudes of the core KDE developers and the general lack of
comments and documentation in so much of the code.

I do not mind "getting my hands dirty" with KMahjongg, but I refuse to
have anything further to do with Frameworks and KF5.  In particular,
I cannot build Frameworks in my system (Apple OS X), so I will not
be able to test anything on the kmahjongg/frameworks branch.

However, I am acquainted with the KMahjongg code and am a reasonably
good player, so I should be able to help you with queries you may have.
And I could at least eyeball/review any changes you make.

> BTW: I find it amazing that you are still active in development. Hats off. :)

Thanks, Frederik… :-)  Check the enhancements to KSudoku, Palapeli
and KGoldrunner in recent and current KDE Applications releases… :-)
These are still KDE 4 apps, of course...

> So:
> 1) frameworks -> master => release ... qgraphics -> master => release
> or
> 2) qgraphics -> frameworks ... frameworks -> master => release
> 
> The polls are open.
> If noboby else has an opinion, I might try to merge qgraphic into 
> frameworks and try out what comes out of it.

I vote for 2), naturally… :-)

All the best, Ian W.

> Regards
> Frederik
> 
>> The long story is here:
>> 
>> https://mail.kde.org/pipermail/kde-games-devel/2015-January/011356.h
>> tml KMahjongg archaeology
>> 
>> In essence, KMahjongg was ported to QGraphicsView, away from the
>> obsolete KGameCanvas graphics, but KMahjongg continues to be
>> released with the obsolete graphics and IMHO it should certainly
>> not go forward into KF5 in that state.  Indeed, I would say that
>> things like KGameCanvas should have been dropped from libkdegames
>> years ago.
>> 
>> I set about doing something about the situation, but this is as far
>> as I got:
>> 
>> https://mail.kde.org/pipermail/kde-games-devel/2015-January/011521.h
>> tml Reviving KMahjongg port to QGraphicsView
>> 
>> The QGV version of KMahjongg ran well - I tested it thoroughly. 
>> Unfortunately it turned out that there were two sets of patches to
>> fix various KMahjongg bugs: some in the QGV branch already and some
>> proposed against the master branch by Jan-Peter Nilsson.
>> 
>> I was attempting to sort out the overlaps and conflicts, when I got
>> diverted into an extended effort of trying to get Frameworks, Qt5
>> and KDE apps working on the Apple OS X platform (or even building,
>> in my case) and I never got back to KMahjongg.  Sorry.
>> 
>> BTW, I am not, nor ever have been, the official maintainer of
>> KMahjongg.
>> 
>> Cheers, Ian W.