[dune-pdelab] Workflow

Markus Blatt markus at dr-blatt.de
Fri Jun 26 18:08:56 CEST 2015


Hi,

all this (including Steffen's workflow) sounds awfully complicated to
me.

On Thu, Jun 25, 2015 at 03:20:11PM +0200, Steffen Müthing wrote:
> Hi,
> 
> 
> > Am 25.06.2015 um 14:10 schrieb Jö Fahlke <jorrit at jorrit.de>:
> > 
> > Am Thu, 25. Jun 2015, 08:58:25 +0000 schrieb Dominic Kempf:
> >> Date: Thu, 25 Jun 2015 08:58:25 +0000
> >> From: Dominic Kempf <gitlab at conan2.iwr.uni-heidelberg.de>
> >> To: jorrit at jorrit.de
> >> Subject: Re: dune-pdelab | [cleanup] remove leftover from mfd code (#28)
> >> 
> >> New comment for Merge Request 28
> >> 
> >> http://conan2.iwr.uni-heidelberg.de/git/pdelab/dune-pdelab/merge_requests/28#note_1156
> >> 
> >> 
> >> Dominic Kempf
> >> 
> >> Oh, well I branched this from an older master. In the meantime, somebody changed a file that I wanted to fully delete. That prohibited a fast forward merge for my first merge request.
> >> 
> >> I solved the problem by cherry-picking my commit onto a fresh master. That resulted in an easy to resolve merge, as I was deleting the file that was having conflicts.
> > 

> That said, I think that merging master into a feature branch should
> be a *last resort* measure;

I agree with this.

> normally, you want to do the exact opposite, i.e. merge feature branches *into* master.
> 
> AFAIU, the default workflow should be:
> 
> 1) See whether the automatic merge using the button works. If it does, use it.
> 2) If it does NOT work, do the following:
> 
> - Push the latest version of your feature branch to GitLab
> - Switch to master - git checkout master
> - Update master to the most recent version - git pull
> - Merge the feature request without fast forward (that can’t happen
> - anyway in this case;

Here my opinion differs:

IMHO contributors should make every effort that their branches can be
merged automatically. Otherwise the maintainers have to do all the
testing twice. I doubt that there is time for this.

If the branch is not mergeable, one should rebase it on top of the current
master and resolve all the conflicts. Yes, this might be a lot of work
and require loads of testing. But it will teach you to keep your pull
requests as small as possible, (Dominic's approach is valid
too: make a new feature branch from  the current master and cherry-pick from
the old from the old feature branch. Amounts to a manual git rebase.)
Now the automatic merge should work again.

Thus one should never have to change the master branch of the forked
repository. It is simply a mirror of the upstream master branch.

Markus

-- 
Do you need more support with DUNE or HPC in general? 

Dr. Markus Blatt - HPC-Simulation-Software & Services http://www.dr-blatt.de
Hans-Bunte-Str. 8-10, 69123 Heidelberg, Germany
Tel.: +49 (0) 160 97590858
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20150626/4cd36ec4/attachment.sig>


More information about the dune-pdelab mailing list