2.6.39 has been uploaded to sid and deprecates -686 in favour of -486 and
-686-pae; the world is not ready for -pae yet, so we now ship -486.

Note that sid's 2.6.39 depends on sid's initramfs-tools. This situation will
probably get easier to deal with for us once this kernel reaches Debian backports.

sid's aircrack-ng needs it.

This reverts commit d3ee66bb.

I was confused, which is not surprising considering the shape of the Broadcom
drivers landscape (http://wiki.debian.org/bcm43xx#Drivers).

The broadcom-sta (wl.ko) driver indeed brings support for some recent WiFi
adapters, but it doesn't content itself with conflicting with the module that
supports the same hardware (brcm80211, brcmsmac): it also conflicts with the ssb
module that is itself needed by a few other modules such as b43, b44. In short,
it seems like wl.ko breaks more hardware support than it brings.

Background info:
  http://wiki.debian.org/brcm80211
  http://wiki.debian.org/wl

User reports concur to indicate that the free brcm80211 / brcmsmac module are
not on par with Broadcom's one, unfortunately.

Installing from unstable for compatibility with newer kernels.

Additionally, the package from unstable automatically blacklists the conflicting
brcm80211 and brcmsmac modules. See Debian bugs #597786, #601009 and #611539 for
details.

Hence remove our home-made backports.
Rationale: those are compatible with the 2.6.38+ kernel we will probably ship
soon.

A fixed kernel now is on security.d.o => no need to play tricks with
proposed-updates anymore.

A fixed kernel now is on security.d.o => no need to play tricks with
proposed-updates anymore.

Debian sid's 2.6.39 has no aufs support (#627573) and is thus not usable for
Debian Live systems as is.

2.6.39 has been uploaded to sid and deprecates -686 in favour of -486 and
-686-pae; the world is not ready for -pae yet, so we now ship -486.

Note that sid's 2.6.39 depends on sid's initramfs-tools. This situation will
probably get easier to deal with for us once this kernel reaches Debian backports.

Linux kernel >=2.6.32-32, that fixes #618665, has landed in there but not in the
main Squeeze repositories yet.

Sounds like 2.6.38 will not land into squeeze-backports early enough.

Some patches that will help us implement our todo/macchanger task have been
integrated in the sid package. This package has made its way into testing
already, but aircrack-ng has not. When aircrack-ng comes back into testing,
we'll probably switch to fetching both of these from there.

This ensures the LD_PRELOAD thing is actually used for most operations during
the build.

Linux kernel >=2.6.32-32, that fixes #618665, has landed in there but not in the
main Squeeze repositories yet.

Some patches that will help us implement our todo/macchanger task have been
integrated in the sid package. This package has made its way into testing
already, but aircrack-ng has not. When aircrack-ng comes back into testing,
we'll probably switch to fetching both of these from there.

This ensures the LD_PRELOAD thing is actually used for most operations during
the build.

This package has not been installed in our Squeeze-based Debian Live images for
weeks.

This package has not been installed in our Squeeze-based Debian Live images for
weeks.

It was removed from Squeeze, and live-boot / live-config might depend on it.
This is a temporary workaround until this is sorted out in Debian Live.

This makes sure our source tree goes on building Squeeze images when Squeeze is
released.

This reverts commit 000c030b.

We *really* want to install live-config* packages explicitly built for Lenny as
they differ e.g. in the way they configure the console keyboard layout.

Fixed live-config* have migrated to Squeeze.
Would be nice to build 0.6 with Squeeze's live-build and live-config
rather than Git snapshots.

A fixed version has entered Squeeze, let's depend on more commonly available
software for our build environment.

Our "normal" pinning installs the versions found in Squeeze.
2.0~a29-1 has migrated to Squeeze.
It changes quite a lot of things (e.g. rename LH_* variables to LB_*)
which we cannot afford dealing with at this stage of the release process.
Indeed upgrading to 2.0~a29-1 would probably imply:

- adapting various bits of our build process (hooks and so on)
- re-doing the testing that has already done on T(A)ILS 0.6~rc1.

Temporary solution: include version 2.0~a21-1 of the live-build and live-helper
packages in our source tree and setup APT pinning so that they are preferred to
the version in Squeeze.

Using Squeeze's live-config does not produce fully-working Lenny images.