converting git to mercurial results in a very strange graph

[Uwe Brauer]

Hi


I am not sure who is the culprit here, but after converting a git repository to mercurial the graph looks like this
--8<---------------cut here---------------start------------->8---

◉ changeset: 7:947d3f4b0d73
tag: default/main
tag: tip
tag: v0.3.0
Branch: main
Author: Wesley Pegden <wpe...@users.noreply.github.com>
Date: Tue, 08 Sep 2026 13:53:03 -0400
Phase: public
Parent: -1:000000000000:default
Summary: Release v0.3.0

○ changeset: 6:30dafdb5a881
tag: v0.2.5
Branch: default
Author: Wesley Pegden <wpe...@users.noreply.github.com>
Date: Mon, 10 Aug 2026 15:19:47 -0400
Phase: public
Parent: -1:000000000000:default
Summary: Release v0.2.5

--8<---------------cut here---------------end--------------->8---



So the nodes are not connected (there are 8 changesets/commits)

When I clone the repository with git and run
git log --graph --topo-order --color=always --all --decorate --pretty | git name-rev --annotate-stdin | less -R


Then I see the same, nodes without connection.

I am really puzzled. The author commited checked out the first commit again changed and commited? Was this the workflow?

[Dr. Arne Babenhauserheide]

Uwe Brauer via Mercurial <merc...@lists.mercurial-scm.org> writes:

> I am really puzzled. The author commited checked out the first commit again changed and commited? Was this the workflow?

Sounds rather like the author created new repositories for each release
and force-pulled them into the main repo.

Maybe a way to implement a multi-branch workflow with git without having
them show up in regular history during development.

(because the starting points of git branches don’t persist in history,
so you can’t see as easily which commits were done as part of release fixes)

Best wishes,
Arne
--
Unpolitisch sein
heißt politisch sein,
ohne es zu merken.
https://www.draketo.de

[Uwe Brauer]

>>> "'ABvh" == 'Dr Arne Babenhauserheide' via hg-git <hg-...@googlegroups.com> writes:

> Uwe Brauer via Mercurial <merc...@lists.mercurial-scm.org> writes:
>> I am really puzzled. The author commited checked out the first commit again changed and commited? Was this the workflow?

> Sounds rather like the author created new repositories for each release
> and force-pulled them into the main repo.

> Maybe a way to implement a multi-branch workflow with git without having
> them show up in regular history during development.

> (because the starting points of git branches don’t persist in history,
> so you can’t see as easily which commits were done as part of release fixes)

Ah, thanks, that makes some sense, although I still consider it as
rather bizarre.