Meeting Note Archival

[iccir]

I'm writing this up as an issue so everybody can opine.

Right now, we use hackmd for meeting notes.

It's great for collaborative editing, but it's not the best from an archival perspective. Features such as full-text search and unlimited history are locked behind membership plans.

As a new member to the team, I struggled to find information covered in old meetings. I also struggled with immense browser slowdown and crashes when viewing/editing meeting notes (I thought it was my computer until Hannah mentioned that others experience it).

I was also confused as to the canonical location for archived notes, as they appeared on both hackmd and this repository.

I'd like the propose the following:

1. We continue to use hackmd for meetings.
2. We make this repo the "truth" for archival purposes. Existing notes on hackmd can redirect to the corresponding location in this repository.
3. We use the "Push to GitHub" feature each month to save the current meeting notes to GitHub. Incoming meeting notes would be saved to a specific path, like incoming/meeting_notes.md or to_import/meeting_notes.md.
4. We use a variation of the tools/import_meeting_notes.py script to extract the meeting notes from the "incoming" path and automatically put them into the corresponding meeting_notes/{year}/…md file.
5. The script keeps the latest X meetings for context (maybe 4 meetings) and saves them back to to_import/meeting_notes.md.
6. The trimmed file can be synced to hackmd via "Pull from GitHub". (Alternatively, we can manually trim the file on the hackmd side - whatever is easier).

Once we get steps 3 and 4 working reliably, it should be possible to either fully automate or at least automatically open up a PR for review.

We could also use the same approach for New Contributors Meetings notes and GSOC meeting notes.

Ideally, I'd like to make it as easy as possible to archive meeting notes so they get saved in searchable form more often.

[story645]

we could probably set up a github action/job to auto sync the notes from this side

[iccir]

we could probably set up a github action/job to auto sync the notes from this side

I'm not sure if it can be fully driven from this side, but we can run a script if we see a commit from the hackmd account.

I'm thinking that if the script encounters any issue (such as an non-parseable # heading or "I don't know what year this meeting corresponds to"), it could open a PR and require a maintainer to sign off on it. Else it could re-commit if no errors are detected.

Ideally, I think the workflow would be:

• Hit "Push to GitHub" in hackmd
• Wait for a confirmation e-mail/notification that the script ran and no errors occurred.
• Hit "Pull from GitHub" to get the updated note for the next meeting.

Assuming we do this once a month for (weekly meeting / new contributors / GSOC), that's a total of 6 push/pulls out of our allocated 20.

[story645]

I'm not sure if it can be fully driven from this side, but we can run a script if we see a commit from the hackmd account.

https://hackmd.io/@docs/Getting-Started-with-the-HackMD-API -> says 400 API calls a month, which I think a weekly scrape of recently updates notes should be fine.

[iccir]

https://hackmd.io/@docs/Getting-Started-with-the-HackMD-API -> says 400 API calls a month, which I think a weekly scrape of recently updates notes should be fine.

The only issue I see is that HackMD doesn't have a concept of a Team API token, nor does it have a concept of fine-grain token control. Hence, any API token created would have access to all of a user's hackmd notes as well as team notes. If we store that token as a GitHub Actions secret, it then extends your personal hackmd account to anyone that can access said secret.

It looks like we have a matplotlib hackmd user account. I guess we would associate the API token with that account?

[story645]

It looks like we have a matplotlib hackmd user account. I guess we would associate the API token with that account?

I think I made that account for those sorts of reasons