Comment 18 for bug 1570472

@Simon, I think the new debdiff is an improvement, but I have the following questions/requests:

1) Can you add more DEP3 headers? Author and Source are missing, I think. I think this is especially important since we're folding multiple upstream commits into one.
2) Can you use dep3changelog to make the changelog? It makes for a more consistent formatting in debian/changelog.
3) I noticed the lib/puppet/provider/service/debian.rb changes from https://github.com/puppetlabs/puppet/commit/7fe61647f23650fc4c93cd6b54c654a20ff7c9f9 do not seem present in your debdiff.
4) From the same commit, I see the change to introduce debian_enabled? has been backported, but no callers have. The caller would be in enabled? which looks quite different in 3.8.5, unfortunately, and might imply more history needs to be imported to properly support systemd? Looking through the commits that came before that one in the enabled? function, one specifically mentions lxc and containers, which feels like a relevant case too.

My primary worry is introduction of regression, potentially, for the SRU. My preference, to avoid that, is for us to match upstream patches exactly, if we are going to fold them together. If we are not going to match exactly, then it seems like (to me) it's better to make individual patches and explain why hunks are dropped in the patch message. This (given an ideal upstream project) also makes it easier to know when something was fixed upstream, and if that upstream fix turns out to be a regression-inducer or have other bugs, it might be so tagged upstream too.

@Robie do you have an opinion here or guidance on the typical process?