# Leaving Redmine for Github Issues

## Motivations

Using pulp.plan.io holds us back:

* Users have to create a pulp.plan.io account to file issues, etc
* Most users already have a Github account
* Github is dominant, by using a boutique tracker Pulp is not-normal
* Also very well understood, people instantly know how to use it, whereas we have to write docs for using redmine.
* "Making people successful in providing feedback, and therefore being part of the project."
* We maintain issue and PR linkages and with both code and issues on github they are integrated
* Spam is still a problem on pulp.plan.io
* pulp.plan.io has gaps in its API and cannot be fully automated. github's API does not have gaps.
* Github syntax conflicts with our redmine syntax (ie it tries to map \#1234 to a PR/issue)
* Github issues is a simpler solution. May help us to simplify tracking fixes/features.

## Obstacles to moving to Github Issues

* What to do about existing issues on redmine? Massive migration?
* Losing history means losing lots of useful information.
* Missing features like custom fields, epics, etc
* e.g. bugzilla links
* redmine-bugzilla automation
* parent tasks are useful
* issue relations are useful
* There is no way to make a comment/issue private
* not that nowadays we are using this a lot, mostly thinking into the future so this does not become a bottleneck
* there are no queries across projects that facilitate triage/sprint planning etc.
* You can see issues for the pulp org: https://github.com/issues?q=user%3Apulp+
* Filter by label: https://github.com/issues?q=archived%3Afalse+user%3Apulp+label%3A%22Needs+Cherry+Pick%22
* there is just opened and closed state - what if we hypothetically have qe in the future, how do we handle on_qa state?
* Would a label work?
* how are epics/subtasks handled?
* Epic tag and link to sub-issues
* Does github have everything to facilitate our sprint/triage/planning processes? Github has kanban support to some extent - github project boards.
* marking as triaged or groomed
* aligning to a certain release or sprint
* putting in a 3-month bucket
* Will have to rely more on labels but contributors outside Pulp org can't edit these.
* Any significant change requires effort and cost with team getting used to new thing
* GitHub projects: workflow sucks for maintainers side, but you can integrate it like Fedora's Taiga instance.
* Migrating notifications would be difficult. Would have to match users.
* overcome obstacle: Github recently added the ability to movie issues within the same org.
* Tracker to downstream: bugzilla (technically I think several other stakeholders link to our redmine but we don't link back)
* Identifying who is our target audience will help us identify best tools?

## Alternatives

###### tags: `PulpCon 2020`