Karmic kernel common debian infrastructure acceptance request

Tim Gardner tim.gardner at canonical.com
Thu Jul 8 14:51:42 UTC 2010


Hello Ubuntu Stable Release Updates Team,

Stefan has informed me that the proposed Karmic upload has not been 
accepted on the grounds that the debian packaging changes are too large 
and do not address a specific bug, thereby violating SRU policy.

As you are no doubt aware there has been an enormous proliferation of 
kernel packages beginning with the netbook branch in Hardy, then rapidly 
multiplying as we added incompatible source code bases for the various 
ARM branches in Jaunty, Karmic, etc. This trend is likely to continue 
until the ARM community can coalesce around a single source code base.

In an effort to reduce the maintenance cost of the various branches I 
started an effort to adopt a common debian packaging infrastructure 
across all releases. Andy has continued that effort and applied it to 
all branches in the Lucid and Maverick releases.

Here is a short synopsis provided by Stefan (which you may have already 
seen on another list):

* Abstracted debian was introduced in the Karmic to allow handling
     the topic (arm, ec2) manageable when needing to include security
     and proposed changes from the master.
* This was SRUed back into Jaunty (and Intrepid) based on our
     verification with debdiff which makes sure that the resulting
     binary packages are the same before and after.
* The downside of the current implementation in Karmic is that
     some files which are used commonly had been split up. So fixing
     a bug in the build system requires the same fix to be done manually
     to all topic branches.
* In Lucid this was changed to make it possible to have common files
     shared so changes would just automatically fall through to all topic
     branches. This mainly consists of moving files around. Though the
     diff seems big.
* Again this has been verified by comparing the binary files before and
     after. The few differences found were actually considered to be bugs
     before and the right results showed up in the new packages (but all
     changes also were not affecting the operation of the kernel, like
     documentation files getting packaged now).
* So the changes are not affecting the binary code and are thought of
     being very useful to have a maintainable build system across all
     topic branches and lessen the risk of errors due to different build
     system in different releases.

In summary I would ask that these proposed Karmic changes be accepted. I 
intend to push for the same changes in Hardy. If necessary we could 
provide a Karmic source package with only the debian infrastructure 
changes, thereby making it easy to compare the outputs from the previous 
version.

Regards,

rtg
-- 
Tim Gardner tim.gardner at canonical.com




More information about the kernel-team mailing list