Re: [PATCH 1/3 v2] rpm: Uprev to rpm-5.4.16 (pre) and rpm-5.4+cvs to current CVS head

[Mark Hatle <mark.hatle@windriver.com>]

On 2/26/16 9:01 AM, alexander.kanavin@linux.intel.com wrote:
>> +Set the DB 6 version to match oe-core db 6.0.30
>> +Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
>> +-    [db-6.1], [db_create], [db.h],
>> ++    [db-6.0], [db_create], [db.h],
> 
> Why not update the oe-core recipe of db to 6.1 instead? Rpm is the only
> thing that is holding it back.

Because today is feature freeze, and I'm not going to introduce a new BerkleyDB
version on the day of feature freeze.  The only thing I can test on is RPM --
and there may be other users.

The patch was done as a single unit specifically so when DB-6.1 is introduced,
it's VERY simple to remove.

> Note that last time I tried, it broke completely when trying to create an
> image, but maybe the problem has been fixed in the rpm upstream meanwhile:
> http://lists.openembedded.org/pipermail/openembedded-core/2015-August/109187.html

There were two issues affecting image generation.  One was an upgrade of the rpm
python module broke the PACKAGEORIGIN support we rely on.  The other was a
processing issue in the package_manager.py.  It was splitting on '|', and if any
of the lines came back as warnings (say from BerkleyDB) it would fall over.

Both of the issues were addressed in this set.

--Mark

> Alex
> 
> 
>