On Fri, Aug 31, 2012 at 09:29:27PM +0200, Philipp Thomas wrote:
On Fri, 31 Aug 2012 13:54:46 +0200, Andreas Jaeger firstname.lastname@example.org wrote:
[ 8< ]
# PATCH-FEATURE-OPENSUSE -- use separate symbol version for Owl extensions - email@example.com
You could generate the patch tag from the data in the patch but not the other way round. Our kernel maintainers have a hard requirement for commented patches, why can't we have that for the rest of the distribution?
As other teams might have found a different but also working approach.
For Samba we ensure to have a header in each patch. By this header we're able to identify the author and the addressed issue.
In the case of an upstream patch this is identical to the output of git show <object> and the name of the object also defines the filename.
For patches we're still working on or which aren't applicable upstream we use an own self describing name but ensure to keep the same patch header.
This allows people pulling the source rpm easily to identify the author and the subject of a particular patch.
But these are all details other teams don't care about. As we don't care much about how the x.org or kernel developers maintain their packages.
Is this the result cause we're all this ignorant?
No, I believe this result is caused by different experiences and needs. And a huge, huge amount of history.
Due to our experience the package change log file has to fulfill different needs:
a) In the case of an update users need to see what's new and got added/ fixed.
b) We reuse the same text to concatenate the update notification messages.