Mailman - investigate long term solutions

[ChristianKuehnel]

goal: investigate alternative solution and help community come to a decision.

This is the long-term perspective to #5

[ChristianKuehnel]

Meeting on 2021-03-30:

• topic is probably to big for the amount of time we have.
• postpone indefinitely

[ChristianKuehnel]

Meeting on 2021-04-13:

• The LLVM Foundation is investigating a potential migration from mailman to
  discourse
• The mailing list archive would be migrated by discourse.org}
  (the company behind the tool)
• The Foundation would like the IWG to manage the migration and investigate
  how to migrate the related workflows.
• we need to understand the workflows on the mailing list and figure out
  how that could be migrated
• use cases we're aware of
  • discussions of community members
  • interaction via email
  • integrations with all the tools sending to the mailing list
• This would be a major effort and require significant time investment from
  the working group members. For this we need explicit buy-in from enough
  members so we can actually manage the work load. Christian will open the
  discussion on the mailing list.

[ChristianKuehnel]

How I would approach the topic (if we decide to do so):

• Discuss with the company how they would migrate the archive and handle GDPR deletion requests.
• Figure out how to interact with discourse purely via email (without using the web UI).
• Scan through the last month (or so) on the mailing lists and figure out what automatic emails there are and how we could move that to discourse (or some other tool).
• Create a mapping of the mailing lists and maybe certain topics (e.g. official announcements, RFCs) to categories in Discourse.
• Create proposals for all workflows that need to be migrated. Figure out which usages of the mailing list will not be possible any more after the migration (e.g. code reviews via email).
• If needed, run a survey with the community to gather more information.
• Figure out steps for the migration and who would have to do that. Figure out which of these we could hand off to a contractor. Come up with a realistic time line for the migration.
• Write all of that into an RFC (or a proposal) and run it by the community to see what we missed.

[ChristianKuehnel]

I started with some initial research into the migration. It looks quite promising so far:
https://github.com/ChristianKuehnel/llvm-iwg/blob/Discourse/Discourse_research.md

I'm keeping the document in my own repo as the working group has not decided if we want to work on that topic.

[sqlbyme]

I was able to get the mbox archives of the lists over to the Discourse folks today. They will have a look this week and get back to us with regards to costs.

[ChristianKuehnel]

The LLVM foundation decided for Discourse as a long-term solution.
See discussion in https://lists.llvm.org/pipermail/llvm-dev/2021-June/150823.html

#47 will track the details of the migration