On Tue, Aug 24, 2004 at 12:13:31PM -0700, Jay O'Brien wrote: > Charles Ulrich wrote: > > > The phrase "ports > > build packages" is a neat and efficient way of rectifying the > > misunderstandings that can occur when trying to give a proper > > explanation of FreeBSD package management.
> Thank you for concisely answering a burning question I didn't > know how to ask. I had gathered the concept that if I built a > "port" I was stuck with it, because it couldn't be removed as > easily as something installed as a package. That phrase -- "ports build packages" is just so right on so many levels. Except for the literal description of what actually happens. If I was feeling pedantic, I'd say "ports build and install software, and also create a packaging list. A package is simply a shortcut method of using that packaging list to gather together all of those files into an archive in order to copy them to a different machine." But it really doesn't have the same ring to it. Most people know that to build a package you simply change to the correct ports directory and type: # make package Which is fine and dandy most of the time. What deserves to be more widely known is that you can create a package fron any already installed port: # pkg_create -b pkg-name which saves a lot of effort trying to recompile things and is a really handy way to back stuff up. As it stands, you can't make a package without also installing the corresponding port. For many reasons it would be very handy to avoid that: ie. to be able to install the port into a chroot'ed area and assemble the package from there. Avoids conflicts with already installed ports or the need to de-install and re-install a port if you want a different set of options for the package. Also a lot of the time it would make it feasible to build packages as a mortal user, rather than requiring root access. This has been discussed on several occasions in the [EMAIL PROTECTED] list, and there are various patch sets floating around. As I understand it though, quite a lot of ports would need remedial work if that functionality was to be introduced, which is what has kept anything from being committed yet. Cheers, Matthew -- Dr Matthew J Seaman MA, D.Phil. 26 The Paddocks Savill Way PGP: http://www.infracaninophile.co.uk/pgpkey Marlow Tel: +44 1628 476614 Bucks., SL7 1TH UK
Description: PGP signature