[oe] [PATCH 00/31] Numerous patches to resolve yocto-compat-layer

Wed Sep 6 23:59:00 UTC 2017

On 9/6/17 6:49 PM, Khem Raj wrote:
> On Wed, Sep 6, 2017 at 4:41 PM, Mark Hatle <mark.hatle at windriver.com> wrote:
>> Current version of these patches, the v2 of 13/31 and the dropping of 29 and 31
>> has been pushed to:
>>
>> git://git.openembedded.org/meta-openembedded-contrib
>> mgh/yp-compat/master-next
>>
> 
> Are these made on top of master-next ?
> if not it would be easier to integrate if they were rebased on top of
> master-next

Yes they are all based on master-next.

--Mark

> 
>> --Mark
>>
>> On 9/6/17 2:22 PM, Mark Hatle wrote:
>>> This patch set attempts to address the 'yocto-compat-layer.py' failures.
>>>
>>> Many of the patches seem like obvious fixes or changes to me.. however some
>>> of them are clearly dubious.
>>>
>>> I would not suggestion treating this as a single take them all or take none,
>>> but as individual changes.. (a few of them may require prior changes...)
>>>
>>> Besides the standard git summary (below), I have attempted to summaries the
>>> individual patches here to help with review:
>>>
>>> [01/31] iproute2: Move tipc enablement to oe-core
>>> [02/31] libml: Move to oe-core
>>> - I will be sending up a corresponding patch to oe-core for these.  See the
>>> other email ( iproute2: Move tipc enablement to oe-core ) for a description
>>> of why this is necessary.
>>>
>>> [03/31] openipmi: Move to meta-networking requires net-snmp
>>> - openipmi in meta-oe requires net-snmp, but meta-oe doesn't depend on
>>> meta-networking.  Easiest resolution was to move it.
>>>
>>> [04/31] glmark2: Set required distro features to opengl
>>> - Should be one of the 'obvious' fixes
>>>
>>> [05/31] openlmi-tools: Only build if the necessary python
>>> - This is one of the dubious patches.  I could not figure out an appropriate
>>> place to 'move' this to that requires meta-python -- as the item itself does
>>> not appear to be appriopriate to put into meta-python.
>>>
>>> [06/31] xorg-drivers: Add opengl distro flag requirement to ati
>>> [07/31] boinc-client: Add distro flag, requires opengl
>>> - Another set of the 'obvious' fixes
>>>
>>> [08/31] networkmanager-openvpn: requires openvpn from meta-networking
>>> - requires openvpn from meta-networking, but meta-oe doesn't depend on
>>> meta-networking.  Easiest resolution was to move it.
>>>
>>> [09/31] telepathy-python: Move to meta-python, requires python-dbus
>>> - This seems like a reasonable move to meta-python
>>>
>>> [10/31] networkmanager: Move to meta-networking, requires dnsmasq
>>> - This moves the rest of networkmanager to meta-networking
>>>
>>> [11/31] python-ldap: Move to meta-networking, requires cyrus-sasl
>>> - I originally was going to move to meta-python, but that doesn't depend on
>>> meta-networking... so move it there instead
>>>
>>> [12/31] python-networkmanager: Move to networking, requires networkmanager
>>> - Due to the prior moves, move this as well to meta-networking
>>>
>>> [13/31] corosync: Requires net-snmp from meta-networking
>>> - Again move to meta-networking due to dep
>>>
>>> [14/31] pywbem: Requires python-m2crypto-native (and others)
>>> - This is one I'm not sure if it's the right thing or not.. but it does
>>> require meta-python.. so move it.
>>>
>>> [15/31] dlm: requires corosync, from meta-networking
>>> - Again move to meta-networking due to prior move
>>>
>>> [16/31] open-vm-tools: only build if meta-networking and meta-filesystems
>>> - open-vm-tools requires libdnet (meta-networking) and fuse (meta-filesystem)
>>> I think this is a dubious way to 'fix' the issue, but I really couldn't think
>>> of a better way.
>>>
>>> [17/31] miniupnpd: Move to meta-networking, requires libnetfilter-conntrack
>>> - Move to meta-networking due to dependency...
>>>
>>> [18/31] enable_gnome.conf: Move the AVAHI_GTK setting
>>> - The yp-compat script does not allow a layer to modify the system wide
>>> configuration (including modifying other recipe behavior).  The AVHAI_GTK
>>> changes the behavior of that recipe, which is detected by the script.  So
>>> I choose to move this to a configuration file that will need to be required
>>> by the user IF they want avahi to use GTK.  I'm not happy with this, but I
>>> don't see any other way to meet the requirements.
>>>
>>> [19/31] glade3: Move PACKAGECONFIG setting to enable_gnome.conf
>>> - Similar to the above, but with PACKAGECONFIG this time.
>>>
>>> [20/31] libart-lgpg: Obsolete, replaced by version in oe-core
>>> - OE core already has this version, remove it.
>>>
>>> [21/31] pinpoint: Add dependency on opengl for cogl
>>> - Add the required distro flag...
>>>
>>> [22/31] vim: Remove xfce vim bbappend
>>> - This breaks the same rule as 18 and 19.  I didn't see a way to 'fix it',
>>> so I just removed the change.
>>>
>>> [23/31] layer.conf: Add meta-networking for the networkmanager
>>> - add meta-networking as a dependency to meta-gnome for networkmanager
>>> this is due to the move above
>>>
>>> [24/31] meta-xfce: depend on meta-multimedia
>>> - There was a duplicate openembedded-layer entry, but I think it was
>>> supposed to be for meta-multimedia.  This resolves a bunch of dependency
>>> errors.
>>>
>>> [25/31] mpv: Requires ffmpeg which has a LICENSE_FLAGS =
>>> [26/31] recipes-multimedia: Add LICENSE_FLAGS
>>> [27/31] recipes-multimedia: Add LICENSE_FLAGS
>>> - Above items all get LICENSE_FLAGS added to match their dependencies.  This
>>> resolves a dependency issue where the required item(s) are not avaialable.
>>>
>>> [28/31] libmtp: remove 'mtp-tools' from PACKAGES
>>> - This is a real bug in mtp-tools, and needs to be fixed.
>>>
>>> [29/31] xfce4-mixer: Remove, gstreamer-0.10 is obsolete
>>> - There is no 'fix' for xfce4-mixer using gstreamer-0.10, so remove it...
>>>
>>> [30/31] florence: update 0.6.3
>>> - Use Andreas Muller's patch from Oct 2016 to git rid of the gstreamer-0.10
>>> dependency.  Resolve the issue with requiring gtk+3.
>>>
>>> [31/31] gstreamer-0.10: Remove as all users are gone and it is
>>> - Remove gstreamer-0.10 as all of the users are now gone.  (This resolves
>>> various preferred provider issues I observed..)
>> --
>> _______________________________________________
>> Openembedded-devel mailing list
>> Openembedded-devel at lists.openembedded.org
>> http://lists.openembedded.org/mailman/listinfo/openembedded-devel