Unlike poor old Humpty Dumpty, Plasma was able to put itself back together again. Over the weekend the poor thing devolved into a pile of mush. Thanks to the amazing dedication and work of the Plasma developers, however, it's back to functional and we can start making forward progress again.
While this page is pretty fun to look at right now, I'm also struck at how much time and energy has gone into that. Having a good API that is consistent and sensible is important for longevity and for third party usage, so it was certainly worth it. That doesn't mean it wasn't a lot of hard running to just end up in the same place functionally, or that breaking other people's plugins is cool.
The goal is to transition libplasma to kdelibs in 4.2, and my personal goal is to prove we can keep both source and binary compatibility while increasing the number of add-ons by another great leap during 4.1. We just can't afford to break the work of others every release; we know that, and so while we bit the bullet this time (something I warned we'd be doing long before 4.0 itself was out even) I look forward to not doing it again for many, many years.
I know Kevin wants to do another round or two of polishing, but post-4.1 I really want to keep it to an absolute minimum of disruption. Such as zero. ;)
Tuesday, April 29, 2008
Subscribe to:
Post Comments (Atom)

3 comments:
How about never putting Plasma in kdelibs? Assuring binary compatibility means you'll eventually be stuck with an outdated API that is just going to sit around using up system memory. I say keep plasma malleable and lean by reserving the option to remove deprecated calls.
I think occasional disruption is important for keeping things fresh. If I make a plasmoid now and it still works in 3 years (without updating it at all), it could displace better alternatives by simple weight of history. Long term binary compatibility breeds bit-rot, in my opinion.
second that.
Being stable (I dare to say even after KDE 5.0!!) is a very good thing.
Hard? Yes, sure!
Will pay off? No doubts!
I do not mind breaking ABI/API compatibility again in 4.1, but 4.2 should really be a stability start point for plasma IMHO.
Post a Comment