[Pkg-utopia-maintainers] Re: Ubuntu's hal changes which are appropriate for Debian

Michael Biebl biebl at debian.org
Wed Apr 18 21:23:20 UTC 2007


Martin Pitt wrote:
> Hello guys,

Hi Martin,

seems my first email got mangled by Thunderbird. So here's a second try.

> yesterday I went through the huge task of merging Ubuntu's and
> Debian's hal (against current experimental svn head). It's quite a bit
> of delta now since it seems that my last set of patches hasn't been
> considered yet.
> 
> I stripped out the Ubuntu specific bits from the diff; here is the
> accompanying changelog part:
> 
>   ** Packaging fixes which should go to Debian **
> 
>   - Privilege reduction:
>     + debian/hal.postinst: Add haldaemon to various groups for removable
>       device access (floppy, plugdev, etc.).
>     + debian/patches/25_privileges-addons.patch: Run addon-storage as
>       non-root.

Definitely needs some closer investigation to not mess something up.
Sjoerd, what's your opinion on this?

Just a quick note: Why don't you remove haldaemon again from the various
groups on purge? Is this intentional?

>   - debian/hal.init.dbus: Use actual init script path in usage help, not
>     hardcoded /etc/init.d/hal (since it is actually in /etc/dbus/event.d).
>     (LP #84642)

For Debian, we decided to get rid of /etc/dbus-1/event.d/. The dbus init
script in unstable already has the required functionality for this.
In the foreseeable future we will transition the start script of
network-manager, dhcdbd, hal etc. to regular sysv init scripts.

>   - debian/hal.postrm: Check mode ($1) properly.
>   - debian/patches/02_powerscripts.patch: Prefer alternative scripts over
>     pmi, since pmi is the default.

In 0.5.9 we have the following ordering:
pm-utils > powersave (*) > pmi > hibernate > /etc/acpi/sleep.sh
(acpi-support) > /usr/sbin/s2ram (uswsusp) > plain /sys/power/state

We don't enforce any particular power management script suite in Debian
(yet) but we want to push pm-utils as preferred solution.
Tim Dykstra and I are already working on it [1] and will upload an
initial version soon.
pm-utils is already the default in Fedora, openSUSE and Mandriva, so
there is a chance for consolidation between the different distros here.
One of the benefits of pm-utils is, that it already uses the quirks
information provided by hal. Atm, acpi-support ships it's own list of
workarounds, the same is true for s2ram etc.
Putting this information in hal would mean, everyone (including other
distros) would benefit.
My plan is to add a Recommends: pm-utils to hal, so it is installed (and
used) by default but leaving open the option for people to use another
solution.

Do you know about Ubuntu's plans regarding pm-utils?


(*) the current unstable branch of powersave (0.15.x) has become a
simple policy agent (similar to g-p-m, just without X depedencies). It
simply calls the HAL methods and doesn't ship its own scripts anymore.
As soon as this version is uploaded to unstable, I'll remove powersave
from this list.
>   - debian/run-hald.sh: Put absolute paths in HALD_RUNNER_PATH so that
>     callouts are found, too.

Agreed, will be applied.

>   - debian/patches/24_ignored_volumes.patch: Ignore partitions without an
>     usable file system (like Mac patch partitions, unformatted ones, etc.)

Agreed, will be applied.

>   - debian/patches/55_nonpolkit-mount-policy.patch: Implement a policy check
>     if PolicyKit is not available: Only allow mounting of removable devices
>     to non-root users.

I already had this patch added for 0.5.9 but removed it in anticipation
of PolicyKit. I'm still a bit undecided, if we should enable PolicyKit
support in hal.
For now it's probably safer to use your patch and disable PolicyKit (and
ConsoleKit) support in hal for now.

>   - debian/patches/62_dbus-python-0.80.patch: Correctly show HAL properties
>     in hal-device-manager with dbus-python. (FD #9343; rejected upstream
>     because h-d-m is going away).

fwiw, I started packaging gnome-device-manager [2], which works also
fine with hal-0.5.9.
So, we can either ship both for now (and using your patch) or already
transition over to the new g-d-m.

>   ** Bug fixes which should go upstream **
> 
>   - debian/hald.8: Remove documentation of removed --retain-privileges
>     option.

As you already noticed, I removed them from SVN already and ship the
upstream man pages now.

>   - debian/patches/16_ntfs_allowed_mount_options.patch: Allow NTFS mount
>     option 'utf8' (LP #78142).

Seems safe, so will be applied, too.

>   - debian/patches/24_ignored_volumes.patch: Ignore fixed partitions but
>     those mounted to /media.

We had a discussion about this sometime ago.
We initially had a patch similar to yours (but more restrictive, it hid
all local partitions), but then users complained that their local
partitions were not shown anymore, so I removed it again.

If we reconsider this decision, I at least would add /mnt to this list.

>   - debian/patches/59_add_ssb_bus.patch: Add support for devices on the SSB
>     bus; patch by Matthew Garrett.
>   - debian/patches/63_fixup_macbookpro.patch: Extend 10-macbookpro-utils.fdi
>     fixes to more MacBooc models; patch by Matthew Garret.

I leave this up to Sjoerd, I really can't comment on that.

>   ** Bug fixes from upstream bugzilla **
> 
>   - debian/patches/65-keyboard-addon-repeated.patch: Do not ignore key
>     repeat events in the keyboard addon. (FD #9767)
>   - debian/patches/66-NULL-fstype-crash.patch: Do not crash on NULL
>     volume.fstypes, and assing some fallback values to them. (FD #10429,
>     LP #87800)
>   - debian/patches/67-fix-probe-volume.patch: Fix scrambled labels.
>     (FD #10362, LP #69914)

Seem to be safe, so I'll add them.

> I took out the diff of debian/control and debian/rules, because it was
> not separatable from the Ubuntu changes. For MacBook [Pro] support,
> you need to add 'pciutils-dev' build dependency and add
> --with-macbookpro and --with-macbook configure options in
> debian/rules.
> 
> 24_ignored_volumes.patch needs a closer look from you as well. In
> Ubuntu we currently ignore partitions which should be 'auto' mounted
> in fstab, but aren't mounted. You probably want to take out this
> snippet from the FDI (I ponder removing it as well, it's a bit painful
> to maintain).
> 
> 02_powerscripts.patch might also be debatable (Debian might have a
> different preference ordering). The other patches should be clear, I
> think.
> 
> I will send the 'Bug fixes which should go upstream' bits to the hal
> ML now for upstream consideration. But you might want to take them
> already.
> 
> Thanks for considering, and have a nice day!
> 
> Martin
> 

Thanks a lot for your efforts. It's really great that you push your
changes back to Debian. To improve this even further, would you be
interested in joining the pkg-utopia group so we can work more closely
together? We could keep the diff smaller this way, which would meand
less work for you in the future.

Cheers,
Michael

[1] http://svn.debian.org/wsvn/collab-maint/ext-maint/pm-utils
[2]
http://svn.debian.org/wsvn/pkg-utopia/packages/experimental/gnome-device-manager
-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 252 bytes
Desc: OpenPGP digital signature
Url : http://lists.alioth.debian.org/pipermail/pkg-utopia-maintainers/attachments/20070418/67d04eb9/signature.pgp


More information about the Pkg-utopia-maintainers mailing list