The ports-system and userland in general.

Raphael Marmier raphael at marmier.net
Tue Dec 14 17:06:20 PST 2004


walt wrote:

If DFly becomes the de facto standard BSD system (which I truly believe
will happen in three to five years) it will attract enough talented ports
maintainers to support any ports system that DFly may elect to support.
Until that day comes, I vote for keeping the FreeBSD ports system.
Add my vote to this.
In the past years quite a few packaging system came up, often with the 
goal of much improving what was before. Despite these efforts, we are 
still at the point that the ports collection is the "less worst" we 
have. And it is one of the older, that should make us sober (and bow to 
jkh) ...

We will definitly not find the magic solution overnight. Therefore we 
should run two efforts in parallel.

- One that defines features we must/should have in an ideal system, and 
try to put it together, without pressure.
- Another one which puts an order of priority on these desired features, 
and tries to add them incrementaly to the current port collection.

Advantages:
- This way, new features are put to test before we end up with a 
complete, elegant system which unfortunatly again is out of touch with 
reality in such subtle ways that it is unusable or at best improves nothing.
- We cannot determine which approach will bring the new package system 
of our dreams, so let's try both at once.

Note that this doesn't not prevent us from fundamentaly altering the 
port's collection's mechanisms, as long as the ports' tree stays 
compatible or convertable by some tool on update. Maybe we should 
dissecate the ports collection, and refactor it, maybe using something 
cleaner than make/Makefiles, but I digress... ;)

my two cents

Raphael





More information about the Users mailing list