Links/scripts to rebuilding next & pu

[ninokierulf]

Link to script to rebuild next based on topics currently in next Link to script to rebuild pu based on topics currently in pu

[ninokierulf]

I have tried to script it, let me know if it's correct or optimizations can be done as I'm not really a bash programmer.

Scenario

• Rebuild pu or next: topics has new changes or rebased, rebuild integration branch

Script

rebuild.sh

target=$1
ancestor=${2:-master}
backup=pre-rebuild-$target

failed="false"
onFail() { failed="true"; echo "Aborting..."; git reset --hard $backup; git tag -d $backup; }
onSuccess() { echo "Rebuild $target success!"; git tag -d $backup; }

# using `mergedinto` alias, this keeps the merge commit order, but removes dups i.e.
# merge A, merge B, then merge A, becomes merge A, B -- in my experience this works well
# https://gist.github.com/rocketraman/1fdc93feb30aa00f6f3a9d7d732102a9#file-gitconfig-L60
git mergedinto $ancestor..$target > /tmp/branches

git checkout $target
git tag -d $backup
git tag "$backup"
git reset --hard $ancestor; echo "Rebuilding $target starting from $ancestor";

# fish shell syntax

while read x; do
  echo "Merging $x"
  # mb (merge-branch) is another alias I use to avoid having to pull all the topic branches down locally
  # but otherwise this is a simple `merge --no-ff`
  # https://gist.github.com/rocketraman/1fdc93feb30aa00f6f3a9d7d732102a9#file-gitconfig-L14
  git mb $x || { onFail; break; }
done < /tmp/branches

if [[ "$failed" == "false" ]]; then onSuccess; fi

Usage:

bash rebuild.sh {integration_branch} {ancestor}

• next: bash rebuild.sh next equivalent to bash rebuild.sh next master
• pu: bash rebuild.sh pu equivalent to bash rebuild.sh pu master

[ninokierulf]

I have another version for graduation where it copies all merged topics from an integration branch to target integration branch

Scenarios:

Copy pu to next Copy next to master

Script

rebuildto.sh

target=$1
from=$2
ancestor=${3:-master}
backup=pre-rebuildto-${target}-from-${from}

failed="false"
onFail() { failed="true"; echo "Aborting..."; git reset --hard $backup; git tag -d $backup; }
onSuccess() { echo "Rebuild $target from $from success!"; git tag -d $backup; }

# using `mergedinto` alias, this keeps the merge commit order, but removes dups i.e.
# merge A, merge B, then merge A, becomes merge A, B -- in my experience this works well
# https://gist.github.com/rocketraman/1fdc93feb30aa00f6f3a9d7d732102a9#file-gitconfig-L60
git mergedinto $ancestor..$from > /tmp/branches

git checkout $target
git tag -d $backup
git tag "$backup"
git reset --hard $ancestor; echo "Rebuilding $from to $target starting from $ancestor";

# fish shell syntax

while read x; do
  echo "Merging $x"
  # mb (merge-branch) is another alias I use to avoid having to pull all the topic branches down locally
  # but otherwise this is a simple `merge --no-ff`
  # https://gist.github.com/rocketraman/1fdc93feb30aa00f6f3a9d7d732102a9#file-gitconfig-L14
  git mb $x || { onFail; break; }
done < /tmp/branches

if [[ "$failed" == "false" ]]; then onSuccess; fi

Usage:

bash rebuildto.sh {target_integration_branch} {from_integration_branch} {ancestor, default: master}

• all pu graduates to next bash rebuildto.sh next pu or bash rebuildto.sh next pu master
• all next graduates to master bash rebuildto.sh master next or bash rebuildto.sh master next master