[Kde-games-devel] Very quiet on kdegames

[Frederik Schwarzer]

Friday 25 November 2011 [Christian Krippendorf]:

Hi

> But there is something other... with git.
> 
> I really cannot believe that all this stuff of migrating to git takes such a 
> big discussion and - more badly - such a big delay. I had read the mailing 
> list about this stuff.
> 
> Sure the binary data data files in git is not that cool as it is handled by 
> other D/VCS but this really coulnd't be a reason, for my opinion. DVCS is and 
> will be the future - don't want to talk about goods and bads of DVCS, this is 
> another story - of all and many many other kde projects are changing to git. 
> All of them had problems, but made i already.
> 
> Having data that blows up the download amount isn't that handy. But first of 
> all DVCS will ever had a more and bigger download size for the first time - i 
> know, binaries are blow this up a bit more - but all in all u will have to 
> download more and more as using DVCS, fur sure just the first time.
> 
> Another thing is: Think of new students or hobby developer that wants to help 
> kde. The first i want to i get the feeling kde isn't well structured. Such a 
> mess, some projects are on git some on svn and so on. This isn't really a good 
> picture of kde at all. Many young developer will first work with git, if 
> someone likes it or not. These people first must be pursuaded into learning an 
> "old" VCS again just for working freely on a project that seem to be not that 
> well structured - as of svn and git mess. And new young developers is the 
> future of the whole kde project.
> 
> There are many more things to say, but this message would get that long... to 
> few time.
> 
> Just think about this is a community project and no solution can make it right 
> to all of us. The migration to git has to get lots of faster as - i think - no 
> one wants to loose the connection.
> 
> btw. i would like to have a separate kdegames-data repo, ... i find it very 
> logic. You can have and work with layouts or tile files without having a game, 
> which maybe is cool for people just designing the layouts and so one. One the 
> other hand kdegames projects will have the package as dependency and will not 
> work without the package, but whatever? kdegames also needs the kde-libs and 
> cannot be run without them. Mainly the same situation from my sight.

There were lots of discussions about the pro and cons of Git and SVN in KDE
land during the last few years. Really long and sometimes opinionated
discussions. But we are beyond that point now. You do not have to convince
us that Git is good. Some of us already know, some know otherwise. It's
pretty much a matter of taste and in general I would say the "DVCS is the
future" statement is not reason enough to switch ...

However, luckily we have past those opinionated times now and someone
already decided where we are going to go. Nice eh? ;)

So, what is holding us back? Simple: manpower, or rather knowledgable
manpower. If we switch, we want to switch properly. The history has to be
consistent and intact and the new repo should be there to stay. So doing
it fast and dirty is not an option.

I migrated some small applications but failed with rule writing for the
kdegames module due to its complexity. Furthermore there was another guy
writing rules for it and had made more progress than me, so I did not
see a point to continue. Well, none of the work-in-progress rules ever
made it to the rules repo, so we are basically back to square one.

In general we as kdegames team do not do the migration. There are svn2git
gods in #kde-git who were doing most of the other modules and I hope they
are kind and sustained enough to get the remaining modules done as well.
But it taked time. Migrating one module can take months from the first
rule file to the final repo.

I am still in favour of separate repositories for each game and keeping
the data files with the source code. The SVG-without-Z approach sounds
pretty nice and I hope that that will be the way to go.

Regards