Files
zooqueenandHanzo Dev 649cac9021 mirror: run in a repository, and ship as a subcommand of the one binary
Reconcile could never have worked. RealGit set no working directory, so every
command ran wherever the process started — for a scheduled job, nowhere — and git
answers "fatal: not a git repository" to the fetch. That error arrives on the
branch that reports GITHUB as unreadable, so a reconciler with perfect
credentials and a perfect table would have called all 24 repos unreachable and
asked a person to look at each one. The dry run passed throughout, because a dry
run resolves names and never touches git.

Git therefore takes the directory as a parameter, and Workspace prepares one
store per repo under a root that MAY survive between runs. May, not must: git's
fetch negotiation offers whatever is already present, so a kept root makes later
runs deltas, and an empty one is equally correct and merely slower. That is what
lets the job use an emptyDir instead of a claim — a ReadWriteOnce volume would
buy the same time at the price of a job pod deadlocking on Multi-Attach whenever
it schedules away from the node holding it.

Both sides are now fetched, not just GitHub. The forge goes first for two
reasons: its tip has to be local before a genuine divergence can be NAMED rather
than reported as "ancestry undetermined", and once its history is present those
objects are offered in negotiation, so the internet side of every run is a delta
even on an empty store. That is the whole reason ten-minute scheduling is
affordable over two dozen repositories.

The proof is a real one. TestFastForwardMovesARealRepository builds two bare
repositories, leaves the forge a commit behind, and requires it to end on
GitHub's tip — then rewinds GitHub so the two genuinely disagree and requires the
forge NOT to move. Every other test in this file uses a fake that agrees with the
code about what git is, which is exactly why none of them caught this. Verified
by mutation: drop `cmd.Dir` and it fails.

Shipping is one binary, not two. cmd/hanzo-mirror becomes cmd/cd-mirror,
dispatched by argv[0] like every other cd command, so the reconciler travels in
the image the control plane already builds, pins and deploys — no second artifact
to version, and no second chance for the two to disagree about which repos exist.
The table travels with it at common.DefaultMirrorTable rather than living in a
ConfigMap, so the JSON that CI validates is the JSON the scheduled run reads.

Co-authored-by: Hanzo Dev <[email protected]>
2026-08-04 16:53:25 -07:00

170 lines
4.8 KiB
Bash
Executable File

#!/bin/sh
#
# Copyright (c) 2006, 2008 Junio C Hamano
#
# The "pre-rebase" hook is run just before "git rebase" starts doing
# its job, and can prevent the command from running by exiting with
# non-zero status.
#
# The hook is called with the following parameters:
#
# $1 -- the upstream the series was forked from.
# $2 -- the branch being rebased (or empty when rebasing the current branch).
#
# This sample shows how to prevent topic branches that are already
# merged to 'next' branch from getting rebased, because allowing it
# would result in rebasing already published history.
publish=next
basebranch="$1"
if test "$#" = 2
then
topic="refs/heads/$2"
else
topic=`git symbolic-ref HEAD` ||
exit 0 ;# we do not interrupt rebasing detached HEAD
fi
case "$topic" in
refs/heads/??/*)
;;
*)
exit 0 ;# we do not interrupt others.
;;
esac
# Now we are dealing with a topic branch being rebased
# on top of master. Is it OK to rebase it?
# Does the topic really exist?
git show-ref -q "$topic" || {
echo >&2 "No such branch $topic"
exit 1
}
# Is topic fully merged to master?
not_in_master=`git rev-list --pretty=oneline ^master "$topic"`
if test -z "$not_in_master"
then
echo >&2 "$topic is fully merged to master; better remove it."
exit 1 ;# we could allow it, but there is no point.
fi
# Is topic ever merged to next? If so you should not be rebasing it.
only_next_1=`git rev-list ^master "^$topic" ${publish} | sort`
only_next_2=`git rev-list ^master ${publish} | sort`
if test "$only_next_1" = "$only_next_2"
then
not_in_topic=`git rev-list "^$topic" master`
if test -z "$not_in_topic"
then
echo >&2 "$topic is already up to date with master"
exit 1 ;# we could allow it, but there is no point.
else
exit 0
fi
else
not_in_next=`git rev-list --pretty=oneline ^${publish} "$topic"`
/usr/bin/perl -e '
my $topic = $ARGV[0];
my $msg = "* $topic has commits already merged to public branch:\n";
my (%not_in_next) = map {
/^([0-9a-f]+) /;
($1 => 1);
} split(/\n/, $ARGV[1]);
for my $elem (map {
/^([0-9a-f]+) (.*)$/;
[$1 => $2];
} split(/\n/, $ARGV[2])) {
if (!exists $not_in_next{$elem->[0]}) {
if ($msg) {
print STDERR $msg;
undef $msg;
}
print STDERR " $elem->[1]\n";
}
}
' "$topic" "$not_in_next" "$not_in_master"
exit 1
fi
<<\DOC_END
This sample hook safeguards topic branches that have been
published from being rewound.
The workflow assumed here is:
* Once a topic branch forks from "master", "master" is never
merged into it again (either directly or indirectly).
* Once a topic branch is fully cooked and merged into "master",
it is deleted. If you need to build on top of it to correct
earlier mistakes, a new topic branch is created by forking at
the tip of the "master". This is not strictly necessary, but
it makes it easier to keep your history simple.
* Whenever you need to test or publish your changes to topic
branches, merge them into "next" branch.
The script, being an example, hardcodes the publish branch name
to be "next", but it is trivial to make it configurable via
$GIT_DIR/config mechanism.
With this workflow, you would want to know:
(1) ... if a topic branch has ever been merged to "next". Young
topic branches can have stupid mistakes you would rather
clean up before publishing, and things that have not been
merged into other branches can be easily rebased without
affecting other people. But once it is published, you would
not want to rewind it.
(2) ... if a topic branch has been fully merged to "master".
Then you can delete it. More importantly, you should not
build on top of it -- other people may already want to
change things related to the topic as patches against your
"master", so if you need further changes, it is better to
fork the topic (perhaps with the same name) afresh from the
tip of "master".
Let's look at this example:
o---o---o---o---o---o---o---o---o---o "next"
/ / / /
/ a---a---b A / /
/ / / /
/ / c---c---c---c B /
/ / / \ /
/ / / b---b C \ /
/ / / / \ /
---o---o---o---o---o---o---o---o---o---o---o "master"
A, B and C are topic branches.
* A has one fix since it was merged up to "next".
* B has finished. It has been fully merged up to "master" and "next",
and is ready to be deleted.
* C has not merged to "next" at all.
We would want to allow C to be rebased, refuse A, and encourage
B to be deleted.
To compute (1):
git rev-list ^master ^topic next
git rev-list ^master next
if these match, topic has not merged in next at all.
To compute (2):
git rev-list master..topic
if this is empty, it is fully merged to "master".
DOC_END