upstream update

1 files changed, 20 insertions, 64 deletions

diff --git a/README b/README
index 4435055b45bc..051685202605 100644
--- a/README
+++ b/README
@@ -1,83 +1,39 @@
 IF YOU ARE IN A HURRY
 =====================
 
-Go to any directory that contains a PKGBUILD and run
+From a directory that contains a PKGBUILD, run:
 
 	arch.build i686
 
-This will build your package for the given architecture. You may also append a
+This will build the package for the given architecture. You may also append a
 list of repositories to use on top of [core], [extra], and [community], for
 instance:
 
 	arch.build x86_64 testing multilib-testing multilib
 
 
-FOR BETTER PERFORMANCES
-=======================
+FOR BETTER PERFORMANCE
+======================
 
-Make sure that hardware virtualization is enabled.
+Enable hardware virtualization.
 
-Make sure that your /tmp is a tmpfs; this is where the copy-on-write drive is
-stored and there is no point that it ever be synced to disk.
+Make /tmp a tmpfs. (This is where the copy-on-write drive is stored.)
 
-Make your host and guest share their pacman package cache through the use of
-the /usr/share/buildstuff/proxy.cgi script; proceed as follows:
-- install any HTTP server on your host computer
-- make this sript run as CGI when called as http://localhost/proxy.cgi
-- make /var/cache/pacman/pkg writeable by the user running this CGI instance
+Make the host and guest share their pacman package cache using the proxy.cgi
+script:
+- install any HTTP server on the host
+- make this sript run as CGI when called as http://10.0.2.2/proxy.cgi
+- make /var/cache/pacman/pkg writeable by this CGI instance
 
 
-HOW THIS ALL WORKS
-==================
+REQUIREMENTS FOR ARCH.INSTALL
+=============================
 
-The arch.install script creates a base Arch Linux disk image, and configures it
-for unattended use in qemu. You may simply call it as:
+The arch.install script creates an Arch Linux disk image which:
+- has [base] and [base-devel] installed
+- has a "user" user able to sudo at will
+- boots unattended with networking enabled
+- runs an SSH daemon accepting the host user's keys
 
-	arch.install i686 ~/vm.img
-
-It creates a sparse disk image and partitions it as 8 GB swap, 64 MB ext2 boot,
-8 GB JFS root, and 16 GB JFS home (where packages are built), and installs it
-with [base] and [base-devel] using syslinux as boot loader and systemd as init.
-All this is to reduce VM latency as much as possible. Eventually the disk image
-amounts roughly to 700 MB of actual disk space.
-
-Any other Arch installer would do too, provided that:
-- the guest system has a "user" user able to sudo at will
-- the guest system runs an SSH server at bootup
-- the host user is able to access the guest "user" user account
-
-The latter is achieved in arch.install by copying the host user's SSH public
-keys into the guest user's authorized_keys file.
-
-Once created, these disk images will be used as read-only most of the time: the
-virtual machines run in arch.build do copy-on-write on top of those base disk
-images, storing the changes into a temporary file discarded after each use. The
-base images are then only updated if they have not been in a day.
-
-Let us go through what happens step by step in arch.build:
-
-1. A (hopefully unique to this instance) random number gets picked: vport; it
-is used as the dmsetup device name, port of the host that gets forwarded to the
-guest SSH's, and name for the temporary copy-on-write drive.
-
-2. If the base disk image does not exist yet, create it by calling arch.install
-
-3. If the base disk image has not been updated for a day, run it within qemu
-and update it. Note that this image never runs any repository beyond [core],
-[extra], and [community].
-
-4. Create the temporary drive and set up the dmsetup copy-on-write instance.
-Run this within qemu.
-
-5. Add to this qemu instance the repositories that the user provided as
-arguments; for instance, above, that was [testing], [multilib-testing], and
-[multilib] so that the former takes precedence over the latter.
-
-6. Update the qemu instance.
-
-7. Run the build job.
-
-8. If everything went well, run cleanup instructions located in cleanup.$vport;
-if anything went wrong, arch.build aborts and it is up to you to kill the qemu
-instance and then run cleanup.$vport after you have investigated the build
-problem.
+Once created, this disk image is read-only most of the time: the virtual
+machines created in arch.build use copy-on-write on top of it.