I appreciate sharing your thoughts in this way. What occurs to me
when reading in particular your 'bp' and 'no-gbp' articles is that
your main concern are "unnecessary" "pre-requirements/dependencies".
You are essentially suggesting to restrict the usage of git to a
transport utility, and also just for the packaging (i.e., the debian/
subdirectory). For the transport of the actual sources, you suggest
to heavily rely on the origtargz utility.
I think you are dismissing that gbp values that packagers can embrace
all functionality that "git" provides. For instance, I value a *lot*
that I don't have to know "quilt" very well. Instead, I can "git am"
patches from configured remote upstream git repostiories. I value that
I can do:
# git format-patch -1 -o debian/patches/new-patch &&
echo new-patch >> debian/patches/series'
I appreciate that I can "git diff --stat" against upstream git branches
to identify what files were stripped or not stripped from the original
upstream sources. I appreciate that I can use 'git grep' to explore
the upstream source code.
Yes, that requires a high level of proficiency with the 'git' tool. It
will not help you as much if upstream doesn't use git. And still, I observe
a momentum towards this workflow, as most recently shown in Ian's work
on 'dgit' and the resulting conversations on debian-devel.