Hi Stippi, Stephan Assmus wrote:
I realize that my voice may not count as I am not a mentor, but allow me to say anyway that I am with PulkoMandy here. For a Haiku GSoC project, I would rather see a student work on improving the native filesystem drivers instead, to make them better, faster and more stable. That would be a well-spent GSoC slot IMO. In all my ignorance, I also fear that if something like LKL would become available, even if as an option, it would remove the incentive from (present and future) developers to do any development for or enhancements to the native filesystem support, and I have a hard time seeing that as a good thing in the long run.There are actually two scenarios. One scenario is where a developer discovers Haiku and finds it so great that he begins work on a feature that he misses. The other scenario, which your argumentation ignores, is that a potential developer discovers Haiku, but removes it from his hard drive and forgets about it soon after, for overwhelming lack of various features. The critical mass is not reached for him. I wonder what happens more often... Judging from how many new blood we manage to acquire, wouldn't you agree it's more often the second scenario?
I actually see a third possible scenario, where unoptimized solutions take hold over native approaches in Haiku, leading in the long run to a system that does not live up to its original claims.
I can see why many people may want things like Qt, KDE, apps like GIMP, etc., LKL and the like on top of Haiku; but AFAIK this was never the way Haiku was intended to be, nor will going down this route promote the development of the good native solutions that will allow Haiku to live up to its claims of being a lean and mean OS that does things differently (better).
IMHO we cannot truely afford the position you take.
I have been hearing this argument for years, but AFAICT, Haiku keeps getting better and both the project, its community and the user base keep growing nevertheless. :)
It's always worth dicussing with these things, since we need to actively prevent bloat, but in this case I belive the project is worthwhile.
I have read Lucian's other emails, and it sounds like his solution would be very well integrated and seamless; while the concern for performance still remain, you might be right after all. :)
Cheers, -- Jorge/aka Koki Website: http://haikuzone.net RSS: http://haikuzone.net/rss.xml
