On Mon, 2012-02-27 at 00:35 +0100, Christoph Wickert wrote:

Hi there,

it doesn't directly affect you, nevertheless I thought I'd let you know that I filed a ticket about muffin and cinnamon in FESCo's trac at https://fedorahosted.org/fesco/ticket/810

I have serious concerns about letting forks like these enter Fedora - not to mention the zombie called Mate. Therefor I think we should clarify our position on forks in general and these ones in particular. Would be nice to get an upstream statement from one of you guys.

Not sure I am ready to give an official 'upstream statement', but I'm happy to state give my opinion.

In general, forking is obviously a vital part of the way our eco system works, and is part of the freedoms guaranteed by the license we choose. Forking an established project allows you to 'stand on the shoulders of giants' and quickly produce something functional, compared to starting from scratch. Just as in nature, most forks end up as unsuccessful experiments that quickly die out and loose interest. Some catch on and either go off into a new direction or eventually supplant the project they forked off from - we all know those examples.

If you are a packager looking at a fork, there's at least 2 considerations:

- There is a lot of duplicated code when a medium-size project is forked completely, and avoiding file conflicts with the original project is probably not on the top of the agenda of the people who started the fork. So you need to be prepared to do quite a bit of work to make the fork install cleanly in parallel to the original project. That includes a lot of busy-work, like changing all the exported APIs to use different names, renaming all libraries, etc etc.

- Since most forks die off quickly, you may want to consider if it is really worth investing that amount of effort - you may end up packaging something that's dead in 6 months...

And maybe one more:

- If the fork you are looking at was done to differentiate another distribution from upstream, then maybe packaging it for your distribution is somewhat counter-productive to the goals of the forkers - they loose their unique selling feature.

So, I guess my message is:

I think packaging forks is fine if you want to scratch your itch, but you are likely to waste a lot of effort and I don't really see how it helps Fedora. If in doubt, I'd rather see you contribute upstream to an existing project such as Xfce, KDE or GNOME. It is a lot of fun !

Matthias