On Fri, 5 Dec, 2025, 11:48 pm Soren Stoutner, <[email protected]> wrote:

> Resending as I realized I had accidentally dropped the debian-python list.
>
> On Friday, December 5, 2025 11:07:44 AM Mountain Standard Time you wrote:
> > On Fri, 5 Dec, 2025, 10:59 pm Soren Stoutner, <[email protected]> wrote:
> > > On Thursday, December 4, 2025 12:45:30 PM Mountain Standard Time Andrey
> > >
> > > Rakhmatullin wrote:
> > > > On Fri, Dec 05, 2025 at 01:12:14AM +0530, Jai kadam wrote:
> > > > >> >> You can also consult its manpage to learn what does it do.
> > > > >> >
> > > > >> >I see, I read the man page just now after your suggestion it's
> not
> > > > >>
> > > > >> detailed.
> > > > >>
> > > > >> It's very detailed
> > > > >
> > > > >By manual you mean "man gbp" or something else?
> > > >
> > > > I mean gbp-push(1), you could learn that it exists from "SEE ALSO" of
> > > > gbp(1).
> > >
> > > To add a bit more information about the behavior of "gbp push” and “git
> > > push”
> > > and when to use them and mix their uses:
> > >
> > > 1.  The key to understanding this is that “gbp push” will push the
> > > following
> > > things:
> > >
> > > a.  The pristine-tar branch (if in use).
> > > b.  The upstream branch.
> > > c.  Any new tags on the upstream branch.
> > > d.  The debian branch, but only if the changelog does not target
> > > UNRELEASED.
> > > e.  Check for a new tag in the debian branch for the latest changelog
> > > entry
> > > and push it, but only if the changelog does not target UNRELEASED.
> > >
> > > 2.  What we want to avoid is pushing a debian/X.Y-Z tag on the debian
> > > branch
> > > before X.Y-Z has actually been accepted into Debian.
> > >
> > > 3.  The following workflow accomplishes this:
> > >
> > > a.  gbp import-orig --uscan (this updates pristine-tar if used, the
> > > upstream
> > > branch, the debian branch, and creates a new upstream tag).
> > > b.  dch  (adjust the changelog version number accordingly, but leave
> the
> > > target as UNRELEASED).
> > > c.  gbp buildpackage (make sure the package builds).
> > > d.  <Fix any problems>.
> > > e.  gbp push (this pushes pristine-tar if in use, the upstream branch,
> and
> > > the
> > > new upstream branch tag, but not the debian branch because the
> changelog
> > > targets UNRELEASED).
> > > f.  git push (this pushes the debian branch).
> >
> > Noted! :)
> >
> > > 4.  When ready to release:
> > >
> > > a.  dch -r (this changes the target from UNRELEASED to unstable, edit
> to
> > > experimental if desired).
> >
> > This is same as dch --release rigth ? Si it also adds the commit messages
>
> Yes, “dch -r” and “dch --release” are the same.
>
> > b.  gbp buildpackage (build the package with the final changelog).
> >
> > > c.  Upload the package.
> > > d.  gbp tag (this creates the debian/X.Y-Z tag, typically one only does
> > > this
> > > after the package has been accepted into Debian).
> > > e.  gbp push (this pushes the debian/X.Y-Z tag).
> >
> > That is use gbp push only once at the very end , right?
>
> No, you also use “gbp push” in step 3.e above.  Otherwise, you need to use
> multiple git commands to do the same things, which would be to push the
> pristine-tar branch (if in use), the upstream branch, and the upstream
> tag.
> If these things are not done at this step, other people who clone the
> repository cannot build the changes you have just made.  When reviewing
> packages that other people have submitted to mentors I do the following.
>
> 1.  gbp clone (if I haven’t already cloned the repository).
> 2.  gbp pull (this pulls all three branches and the tags, so that I can
> review
> the most recent changes).
> 3.  gbp buildpackage.
>
> If you don’t get step 3.e and 3.f right, "gbp buildpacakge" will fail for
> me.
>
Oh i see cause debian-branch will have the previous version of code cause
gbp never uploads it while in unreleased state and we only do dch -r when
we get a green flag from a mentor or maintianer of the package which then
sets it to unstable and then when we run gbp tag and gbp push all work is
done.

Is my understanding rigth ?

> --
> Soren Stoutner
> [email protected]
>

Reply via email to