On Mon, Jan 28, 2013 at 07:37:39PM +0100, Julien Cristau wrote: > Hi Gonéri,
Hi Julien,
> > +
> > +case "$1" in
> > + configure)
> > + [ -d "/etc/fusioninventory" ] || mkdir "/etc/fusioninventory"
>
> That's going to depend on umask. Can the package deal with whatever
> permissions this can end up with?
The software run with root privilege, so believe it won't hurt.
> > + ucf /usr/share/fusioninventory/etc/agent.cfg
> > /etc/fusioninventory/agent.cfg
> > + ucfr fusioninventory-agent /etc/fusioninventory/agent.cfg
> > +esac
> > +
> > +exit 0
> > only in patch2:
> > unchanged:
> > --- fusioninventory-agent-2.2.3.orig/debian/tools/prepare-bpo.pl
> > +++ fusioninventory-agent-2.2.3/debian/tools/prepare-bpo.pl
>
> more noise :(
Indeed, this script should not be here.
> > ---
> > fusioninventory-agent-2.2.3.orig/debian/patches/upstream-fix-backport.diff
> > +++ fusioninventory-agent-2.2.3/debian/patches/upstream-fix-backport.diff
>
> what specifically does that fix? I'm not keen on reading a 1000 line
> diff without knowing what it's supposed to do. At this stage we should
> only be fixing RC bugs, not random other stuff.
This patch brings fixes from the current 2.2.x stable branch of
fusioninventory-agent. As I said, it only includes fixes. Admittedly not
all of there would be RC.
May problem is, if I drop these patches, the 2.2.3 release included in
Debian will way to buggy in my opinion. If I decided to follow Debian
logic and only keep the changes that deserve a RC, I will have to long term
support on a FusionInventory Agent fork. That's something I would avoid if
possible.
I don't want to force the release team to follow my logic, and I will gladly
follow your decision.
This is the simplied Changes about this patch.
General:
* additional hack for LG screens (#1848)
* additional model for ACER screens hack (#1840)
* fix directory creation error handling with older File::Path versions (#1817)
* fix initial delay randomness (#1809)
* fix spurious null character in decoded strings (#1837)
* better fix for size parsing with HP RAID controllers, avoiding warnings
(#1807)
* implement model-specific hack for ACER screens serials (#1607)
* Fix: deal with broken last_state file
* Fix: fix HP RAID size parsing
* Fix: virtual machine inventory with VMware desktop
* Fix: add /usr/local/{bin,sbin} in default $PATH
* Fix: non blocking flock() on log file
Linux:
* fix parsing of newer ifconfig output, such as used on Fedora 17
* Fix: collect qemu -drive information (Alexander Evseev)
Best regards,
--
Gonéri Le Bouder
pgpH9C5YgTu_j.pgp
Description: PGP signature
