On Mon, August 15, 2005 10:25 am, Andreas Hauser said: > I doubt that we will ever have the manpower to maintain our own big > package system unless we we use a parasitic approach in which we > take well maintained ones, like ports, modify them and keep some > way to sync. Let's call it metaports. [snip] > Committing stuff to pkgsrc will still be your biggest problem, since > it looks like you will have maybe one committer (joerg), who is already > packed enough. That means you still might want to apply the above
1 committer is more than what we have with FreeBSD ports; as far as I know, the FreeBSD project does not want to modify the way ports work just to accomodate DragonFly. We have had individual port maintainers able to incorporate fixes suggested by DragonFly developers, in some cases, but that's on a per-port basis, and certainly doesn't hold for all ports or the overall framework. Ports worked well on earlier versions of DragonFly, but as the trees diverge, more ports will break because of DragonFly changes. I also expect FreeBSD to focus more and more on versions 5+, whereas we are compatible with FreeBSD 4.x. Binary compatibility will break with the next DragonFly release. We will always be playing catch-up, and we will _always_ have a poorer version of ports than what FreeBSD has. Here's some positivity, not negativity: the pkgsrc project is designed to run on more platforms. DragonFly-specific features can be added, too, such as: http://www.dragonflybsd.org/docs/USENIX2005_BSD/img4.html Having DragonFly as part of pkgsrc means that it'll be worked on by other pkgsrc developers, and included as part of the quarterly builds. Having something in pkgsrc means that it will generally work for DragonFly.
