BAD_RECOMMENDATIONS is broken

BAD_RECOMMENDATIONS is broken

[Sebastian Spaeth]

minimal.conf has
DISTRO_EXTRA_RRECOMMENDS += "[...] avahi-daemon avahi-autoipd [...]"
which leads to task-distro-base recommending those packages.

Trying to override this with
BAD_RECOMMENDATIONS="avahi-daemon avahi-autoipd"
does not wotk though, it's still pulled in.

This is what rootfs_ipk.bbclass does:

        STATUS=${IMAGE_ROOTFS}${libdir}/opkg/status
        # prime the status file with bits that we don't want
        for i in ${BAD_RECOMMENDATIONS}; do
                echo "Package: $i" >> $STATUS
                echo "Architecture: ${TARGET_ARCH}" >> $STATUS
                echo "Status: deinstall ok not-installed" >> $STATUS
                echo >> $STATUS
        done

which actually writes out the correct stuff

+STATUS=/home/[...]/tmp/rootfs/shr-image/usr/lib/opkg/status
+for i in avahi-daemon avahi-autoipd
+echo 'Package: avahi-daemon'
+echo 'Architecture: arm'
+echo 'Status: deinstall ok not-installed'
+echo
+for i in avahi-daemon avahi-autoipd
+echo 'Package: avahi-autoipd'
+echo 'Architecture: arm'
+echo 'Status: deinstall ok not-installed'
+echo

However, when task-distro-base is installed it still pulls in
avahi-daemon, so I think this way of handling BAD_RECOMMENDATIONS is
broken by different opkg behavior.

Any solution (besides removing avahi from minimal.conf where I don't
think it should be in the first place)?

spaetz

Re: BAD_RECOMMENDATIONS is broken (opkg issue)

[Sebastian Spaeth]

additional: as this is an opkg issue, I am using opkg-native r240.

Graham, is this supposed to work?

Thanks,
Spaetz

Re: BAD_RECOMMENDATIONS is broken (opkg issue)

[Graham Gower]

2009/11/18 Sebastian Spaeth <Sebastian@sspaeth.de>:
> additional: as this is an opkg issue, I am using opkg-native r240.
>
> Graham, is this supposed to work?
>
> Thanks,
> Spaetz
>

A quick look at the relevant code for this indicates to me that it may
work under certain circumstances. Its non trivial to determine how
this all interacts from code inspection alone. I'll play with it for a
bit and let you know what I come up with.


-Graham

Re: BAD_RECOMMENDATIONS is broken (opkg issue)

[Phil Blundell]

On Thu, 2009-11-19 at 07:50 +1030, Graham Gower wrote:
> 2009/11/18 Sebastian Spaeth <Sebastian@sspaeth.de>:
> > additional: as this is an opkg issue, I am using opkg-native r240.
> >
> > Graham, is this supposed to work?
> >
> > Thanks,
> > Spaetz
> >
> 
> A quick look at the relevant code for this indicates to me that it may
> work under certain circumstances. Its non trivial to determine how
> this all interacts from code inspection alone. I'll play with it for a
> bit and let you know what I come up with.

It's meant to be handled by this chunk of code, and I'm fairly sure that
this used to work reliably.  Has that part of opkg changed recently?

                    /* user request overrides package recommendation */
                    if (satisfying_pkg != NULL
                        && (compound_depend->type == RECOMMEND ||
compound_depend->type == SUGGEST)
                        && (satisfying_pkg->state_want == SW_DEINSTALL
|| satisfying_pkg->state_want == SW_PURGE)) {
                         opkg_message (conf, OPKG_NOTICE, "%s: ignoring
recommendation for %s at user request\n",
                                       pkg->name, satisfying_pkg->name);
                         continue;
                    }

p.

Re: BAD_RECOMMENDATIONS is broken (opkg issue)

[Graham Gower]

2009/11/19 Phil Blundell <philb@gnu.org>:
>
> It's meant to be handled by this chunk of code, and I'm fairly sure that
> this used to work reliably.  Has that part of opkg changed recently?

Nope, no changes there. And no particularly invasive changes up to r240.

I have some minor worries that
http://code.google.com/p/opkg/source/detail?r=328 will have some non
obvious fallout. (Obviously not the issue here though)

-Graham

Re: BAD_RECOMMENDATIONS is broken (opkg issue)

[Denys Dmytriyenko]

On Thu, Nov 19, 2009 at 09:25:06AM +1030, Graham Gower wrote:
> 2009/11/19 Phil Blundell <philb@gnu.org>:
> >
> > It's meant to be handled by this chunk of code, and I'm fairly sure that
> > this used to work reliably.  Has that part of opkg changed recently?
> 
> Nope, no changes there. And no particularly invasive changes up to r240.
> 
> I have some minor worries that
> http://code.google.com/p/opkg/source/detail?r=328 will have some non
> obvious fallout. (Obviously not the issue here though)

FYI, http://thread.gmane.org/gmane.comp.handhelds.openembedded/30417

-- 
Denys