Automatically create new milestone; optionally move issues to it

[anthonyfok]

Last Thursday, July 29, Joost and I had a brief discussion on the possibility of automatically creating a new milestone and moving issues to it, and how I surmised that Hugo developers might be already doing it. The day after though, I was surprised by the following tweet:

https://twitter.com/GoHugoIO/status/1421181681717944323

Dear @github moving 200+ issues from a milestone to another is a surprisingly time consuming process. /cc @natfriedman
(and yes, we have been wining about this before)

So, the Hugo folks were migrating issues to new milestone manually too? 🤔

[Edited 2021-08-10: Probably through GitHub's web interface, e.g. search for is:open is:issue milestone:v0.87, click the "select all" checkbox, then click the Milestone to set a new milestone, e.g. v0.88. Problem is that the "select all" checkbox only selects up to 25 issues while Hugo has 299 open issues. 300 ÷ 25 = 12, i.e. the operation has to be repeated 12 times!?! That is a whole lot of clicking! Maybe that is what the complaint was about.]

Anyhow, GitHub does have an Issues API that can create a milestone and update an issue, for example, so perhaps we can use the Issues API to accomplish what we need.

Some on-line posts even suggests the use of GitHub CLI (gh) to manage milestones · Issue #1200 · cli/cli.

(Aside: So that was what I started doing last Saturday (I had a memory lapse), but veered off to looking into having github-cli (gh) packaged for Debian.)

Update: Shared the idea with Hugo at https://discourse.gohugo.io/t/script-to-move-issues-to-new-milestone-proof-of-concept/34181

[anthonyfok]

This is a test script that successfully move OpenDRR/opendrr-api "Sprint 39" issues to "Sprint 40" milestone. (Tested on Thu 2021-08-05).

Note:

1. It lives on my computer as ~/bin/milestone-test.sh
2. GitHub CLI gh automatically replaces {owner} and {repo} based on the Git repo directory that you are in.
3. This is very preliminary and a proof of concept. The milestones names are hardcoded. No error checking is done whatsoever, at least not yet. Proceed with caution!

#!/bin/bash

set -x

OLD_MILESTONE_TITLE="Sprint 39"
NEW_MILESTONE_TITLE="Sprint 40"

OLD_MILESTONE_NUMBER=$(gh api -X GET "repos/{owner}/{repo}/milestones" --jq ".[] | select(.title == \"$OLD_MILESTONE_TITLE\").number")
NEW_MILESTONE_NUMBER=$(gh api -X GET "repos/{owner}/{repo}/milestones" --jq ".[] | select(.title == \"$NEW_MILESTONE_TITLE\").number")

# Print open issues in the previous milestone 
gh api -X GET "repos/{owner}/{repo}/issues" -f milestone="$OLD_MILESTONE_NUMBER" --jq '.[] | [.number, .title, .assignee.login]'

# Move issues to the new milestone
for i in $(gh api -X GET "repos/{owner}/{repo}/issues" -f milestone="$OLD_MILESTONE_NUMBER" -f direction=asc --jq '.[].number'); do
  echo "Moving Issue $i from \"$OLD_MILESTONE_TITLE\" to \"$NEW_MILESTONE_TITLE\"..."
  gh api -X GET "repos/{owner}/{repo}/issues/$i" > "Issue-${i}-${OLD_MILESTONE_TITLE// /-}.json"
  gh api -X PATCH "repos/{owner}/{repo}/issues/$i" -f milestone="$NEW_MILESTONE_NUMBER" > "Issue-${i}-${NEW_MILESTONE_TITLE// /-}.json"
done

To create a milestone:

gh api -X POST repos/{owner}/{repo}/milestones -f title="Sprint 40" -f due_on=2021-08-26T15:30:00-07:00

• GitHub stores only the date of due_on, but when creating a milestone with -X POST, it needs the whole date+time+timezone string like the above.
• No such restriction for changing the due_on with -X PATCH though. For example:

gh api -X PATCH repos/{owner}/{repo}/milestones/35 -f due_on=2021-08-05

TODO:

• Expand this script and turn it into something useful that our team can use. Add features such as:
  • List all open assigned issues
  • Create new milestone with due date
• Make it cross-platform, i.e. can use on Linux, macOS and Windows. (Implies rewriting in Python? Or keep the bash script but provides also .bat / .ps1?)

[anthonyfok]

My quick-and-dirty one-liner that I used today for creating Sprint 44. Perhaps make it a built-in (hardcoded) or user-configurable list?

for i in boundaries data model-factory opendrr-api opendrr-data-store opendrr-platform pygeoapi; do
    ( cd $i && pwd && gh milestone "Sprint 43" "Sprint 44" )
    echo; echo
done

Another TODO: Remove jq dependency and rely on gh's built-in jq functionality alone.