Date: Mon, 30 Jan 2012 23:50:52 +0000 From: Matthew Seaman <m.seaman@infracaninophile.co.uk> To: freebsd-questions@freebsd.org Subject: Re: Unable to upgrade packages on FreeBSD Message-ID: <4F272CDC.5010408@infracaninophile.co.uk> In-Reply-To: <CAGy-%2Bi9pYgB3VjG8KQg98Bfr5Ax2BOLOnuqrzOe_P5juDe%2BVjw@mail.gmail.com> References: <CAGy-%2Bi-6GLfoUuhUExjnVEKhM00TuUimhKuhboLkjBeXNk9hFg@mail.gmail.com> <20120130215826.140fa9df@mpw> <CAGy-%2Bi9pYgB3VjG8KQg98Bfr5Ax2BOLOnuqrzOe_P5juDe%2BVjw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enig25BCD70EE623D97C543F68D9
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
On 30/01/2012 22:04, David Jackson wrote:
> Binary packages are a big time saver and are more efficient.
Yes, definitely -- this is true for many use cases. Not all by any
means, but enough that binary packages are a must-have.
> It should be easy for FreeBSD to make it easy to install the most
> recent versions of all binary packages, its beyond belief they cannot
> pull off such a simple ans straight forward, and basic part of any
> OS.
Now this I dispute absolutely. Whatever gave you the idea that
generating and maintaining an archive of binary packages was at all
"simple and straight forward?" It is most emphatically neither of those
things.
Firstly there's a matter of the scale of the job -- the ports contains
around 23,000 different software packages. That's pretty respectable
compared to most Linux distributions, remembering that there are several
hundred packages' worth of stuff in the base system which would have to
be packaged in a comparable Linux system. Most of those software
packages are under active development, and virtually none of them are
prepared to alter their release schedules one iota to suit FreeBSD.
Just keeping that collection current is a huge task, let alone trying to
maintain and improve the system used to do it.
Then there's the small matter of compiling all that software to produce
the binary packages. At the moment there are 3 different major OS
versions supported across two Tier-1 architectures (i386, amd64 --
everything is expected to work on Tier-1) and four Tier-2 architectures
(ia64, sparc64, powerpc, pc98 -- which should be supported for package
building, but only on a 'best efforts' basis) plus maybe 3 or 4 other
experimental architectures like arm and mips which have the potential to
become very important in the future as they are the basis of a lot of
embedded computing devices. And people have the temerity to complain if
updates aren't available online within a few days!
To support all that takes some pretty impressive computing power spread
over three different data centers (I believe), all of which has been
*donated* to the FreeBSD project, and all of the power, cooling,
bandwidth, maintenance and other ongoing hosting costs are similarly
supplied by donation. Not to mention a hard-core of about 20 key ports
committers, plus maybe a hundred-odd other committers taking a more
peripheral role, and some 4,000 other volunteers that do the work of
maintaining everything.
All of this elides one of the insanely great features of the ports --
which is how configurable and adaptable they are. The trouble is, the
design of the ports really does work best for compiling from source.
There is functionality there which is somewhere between "incredibly
difficult" and "simply impossible" to push up to a set of pre-compiled
binary packages. (Which, by the way, is a feature common to all binary
packaging systems: you always get whatever someone else thought was a
good idea at the time.)
The ports really are one of FreeBSD's crown jewels, and as a system for
compiling software from source and installing and maintaining the
results it has few peers. It is certainly true that FreeBSD's binary
package management could be better. Binary package management under
FreeBSD has always been seen as bit of a second choice compared to
ports, and consequently it has not had the same sort of development
effort put into it. Until recently, that is. We have literally just
had the announcement of the beta test version of the new next-generation
binary packaging system on the freebsd-ports@... list earlier today.
Don't get too excited though -- it will be months at the very least
before pkgng goes into anything like production.
Cheers,
Matthew
--=20
Dr Matthew J Seaman MA, D.Phil. 7 Priory Courtyard
Flat 3
PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate
JID: matthew@infracaninophile.co.uk Kent, CT11 9PW
--------------enig25BCD70EE623D97C543F68D9
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.16 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk8nLOQACgkQ8Mjk52CukIwbkACfShcW8pp2anxNlUaQhxFLWuWZ
MYMAoIBaBhWnCDsy6CQqZtjg+9oXMOaI
=btX3
-----END PGP SIGNATURE-----
--------------enig25BCD70EE623D97C543F68D9--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4F272CDC.5010408>