magic-folder development roadmap?

[David Stainton]

Dear Brian Warner,

Would you have time to help us figure out the most suitable
development roadmap for the Tahoe-LAFS Magic-Folder project?

I believe that at this time we need your senior developer experience
to help us make some decisions. Without this kind of leadership we
will continue to be stuck in Zeno's paradox where we make progress but
there is always more to do.

Lately Daira, Meejah and I have been adding minor features and fixing
minor bugs... however I believe that these are not necessarily the
best things to work on right now.

When and if you have time to chat about this project, here are some of
the things we'd like to discuss with you:

We could stick with our current design which would only have conflict
detection for 2 parties. But then we'd have two additional design
flaws IMHO:
 - ambient authority among clients to perform file deletions
 - an "admin" role with excess authority

To fix our conflict detection we could implement Leif's or Zooko's proposal:
https://github.com/tahoe-lafs/tahoe-lafs/blob/2438.magic-folder-stable.8/docs/proposed/magic-folder/multi-party-conflict-detection.rst

Leif's design preserves history and Zooko's design allows for file
deletions. I could see a need for either design... though they have
different use cases. I would like us to implement Leif's design
because fits my use case. If we do this then we'd close these tickets:

https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2582
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2565
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2551

Leif's design preserves all file history and thus eliminates the
possibility of clients removing file objects forever. Additionally
Leif's design eliminates the "administrator" and it's excess authority
from our original design by implementing a subscription model and
resolving this ticket:
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2589


Sincerely,

David