Bruno Haible via Bug reports for the GNU Internet utilities <bug-inetutils@gnu.org> writes:
> Simon Josefsson wrote: >> So I'd prefer to say ... >> that merge requests on codeberg/gitlab are fine too. > > I will be interested in hearing about your experience with merge requests. I'm rather skeptical to them, but I want to expose myself to using it to gain an informed opinion. As long as it doesn't break existing workflows I see no harm in doing so. > What I can see so far, in comparison with posting a patch, is that > codeberg omits the information about the committer and the commit date: > In > https://gitweb.git.savannah.gnu.org/gitweb/?p=inetutils.git;a=commitdiff;h=f6b14c9afeb0baedf235303545b05833ea497ce6 > it looks like Collin pushed his patch himself, and already two weeks ago. > There is no trace of the fact that you approved and committed it, > nor of the date when you did so. (In fact the commit is identical to > the one Collin submitted at > https://codeberg.org/inetutils/inetutils/pulls/5/commits .) > > Maybe the advantages are worth this drawback? Maybe that should be an RFE > w.r.t. Foregejo? Interesting, I hadn't thought about this. I think most codeberg/github/gitlab projects and folks doesn't see this because the defaults are to use merge commits, which would record this kind of information in a different way. I think merge commits are really ugly so I disallowed them on the inetutils project. So either we change workflow and start to use merge commits (I'm not a fan of this right now but open to change) or we see if we improve things without using merge commits. I think that if I hadn't used the codeberg web interface to approve the merge request, but instead pulled it locally and merged it myself, I could have used some git command and parameter that would result in the git commit identity would point to me instead. We can try this with the next merge request that comes along. So maybe this is a pragmatic way forward. I'm not sure about git signatures here though. What if Collin PGP/SSH signed the commit, would it be visible in this scenario, or would my PGP/SSH signature be used instead? I think local git merge is the only way to get my PGP/SSH git commit signature applied, I'm not aware of any way for codeberg's web interface to sign the commit on my behalf. I agree we could consider a RFE on forgejo on this. What would we want to actually happen here? That clicking the 'approve' button would set the git commit identity to my codeberg user account? Optionally add a Signed-Off: header? Another approach may be for me to pull the merge request locally, amend the commit with a Signed-Off: header (and PGP sign it), and then force push it back to the codeberg merge request, and then approve that? I think establishing a reasonable workflow here will take some time... /Simon
signature.asc
Description: PGP signature
