Hi all,
This affects anyone who has pulled, cloned or forked orGUI in the time frame of
Affected dates: 2026-07-27 - 2026-08-11
If you have a local clone or a fork of orGUI, please do not run git pull
until you have followed the steps below. A plain pull will merge the old and new
history together and duplicate 52 commits.
| when |
what |
| 2026-07-27 21:11 -04:00 |
7459b38 authored |
| 2026-07-27 21:15 -04:00 |
pushed to reciprocal_maps — trailer first public |
| 2026-08-04 21:34 UTC |
#69 merged, so it reached master |
| 2026-08-11 |
rewritten and force-pushed; trailer removed |
The commit reached master through #69 rather than directly, which is why a
commit from late July affects branches merged after it.
What happened
I had to force push to change a commit message. No code was changed
That is also why the steps below are safe: a rebase of your own work onto the
new history cannot produce a content conflict from the rewrite itself.
Affected branches
| branch |
old tip |
new tip |
master |
d2421e1 |
98b3e69 |
reciprocal_maps |
a37bbcf |
ab37f5d |
DWBA |
d2b7fb1 |
33aa9e9 |
esrf_id31_SIXC_Q_version |
d098e06 |
a67b7b5 |
fix-ctr-interface-layer-accumulation |
f2aa65c |
2d7fa99 |
fix/issue-23-db-io-hardening |
5d71e1c |
db2db16 |
No other branch or tag was touched.
Step 0 — back up and check what you have
Commit or stash any uncommitted work first, then:
git fetch origin --prune
git branch backup/pre-rewrite-master master
Now check whether you have local commits that were never pushed:
git log --oneline d2421e1..master
- No output → you have no local work. Use Path A.
- Some commits listed → those are yours. Use Path B.
Path A — no local work of your own
git checkout master
git reset --hard origin/master
If you also have any of the other affected branches checked out locally:
for b in reciprocal_maps DWBA esrf_id31_SIXC_Q_version fix-ctr-interface-layer-accumulation fix/issue-23-db-io-hardening; do
git rev-parse --verify -q "$b" >/dev/null && git branch -f "$b" "origin/$b"
done
Path B — you have your own commits
Replay your work onto the new history. The second argument is the old tip
from the table above:
git rebase --onto origin/master d2421e1 master
For a feature branch that forked off one of the other rewritten branches, use
that branch's old tip — for example, a branch off DWBA:
git rebase --onto origin/DWBA d2b7fb1 my-feature
Do not use git merge-base to find your fork point here. After the
rewrite it resolves back to the commit before 7459b38, which would replay
all 52 rewritten commits as duplicates. Use the explicit old SHA.
If a rebase does stop on a conflict, it is a genuine collision between your work
and someone else's — not an artefact of the rewrite, which is message-only.
Step 3 — verify
git log --oneline -1
git status -sb
You should see 98b3e69 (or the current master commit sha) at the tip of master, no divergence from
origin/master, and 0 trailers.
Then clean up your safety branch:
git branch -D backup/pre-rewrite-master
If you have a fork on GitHub
Do the same in your local clone of the fork, then force-push it to bring your
fork's copy in line:
git push --force-with-lease origin master
Merged pull requests
#65, #66, #67, #68 and #69 have a merge commit that moved. They remain merged
and their diffs still display correctly; only the "merged commit … into master"
link points at a commit no longer reachable from master. Each PR has a comment
recording its old and new SHA. Nothing to do here.
Please comment on this issue if you hit anything the steps above do not cover.
Hi all,
This affects anyone who has pulled, cloned or forked orGUI in the time frame of
Affected dates: 2026-07-27 - 2026-08-11
If you have a local clone or a fork of orGUI, please do not run
git pulluntil you have followed the steps below. A plain pull will merge the old and new
history together and duplicate 52 commits.
7459b38authoredreciprocal_maps— trailer first publicmasterThe commit reached
masterthrough #69 rather than directly, which is why acommit from late July affects branches merged after it.
What happened
I had to force push to change a commit message. No code was changed
That is also why the steps below are safe: a rebase of your own work onto the
new history cannot produce a content conflict from the rewrite itself.
Affected branches
masterd2421e198b3e69reciprocal_mapsa37bbcfab37f5dDWBAd2b7fb133aa9e9esrf_id31_SIXC_Q_versiond098e06a67b7b5fix-ctr-interface-layer-accumulationf2aa65c2d7fa99fix/issue-23-db-io-hardening5d71e1cdb2db16No other branch or tag was touched.
Step 0 — back up and check what you have
Commit or stash any uncommitted work first, then:
Now check whether you have local commits that were never pushed:
Path A — no local work of your own
If you also have any of the other affected branches checked out locally:
Path B — you have your own commits
Replay your work onto the new history. The second argument is the old tip
from the table above:
For a feature branch that forked off one of the other rewritten branches, use
that branch's old tip — for example, a branch off
DWBA:If a rebase does stop on a conflict, it is a genuine collision between your work
and someone else's — not an artefact of the rewrite, which is message-only.
Step 3 — verify
You should see
98b3e69(or the current master commit sha) at the tip ofmaster, no divergence fromorigin/master, and0trailers.Then clean up your safety branch:
If you have a fork on GitHub
Do the same in your local clone of the fork, then force-push it to bring your
fork's copy in line:
Merged pull requests
#65, #66, #67, #68 and #69 have a merge commit that moved. They remain merged
and their diffs still display correctly; only the "merged commit … into master"
link points at a commit no longer reachable from
master. Each PR has a commentrecording its old and new SHA. Nothing to do here.
Please comment on this issue if you hit anything the steps above do not cover.