duplicity-team team mailing list archive

[Stefan Breunig <stefan-launchpad@xxxxxxxxxxx>]

1.
Sure, but the product is called "Amazon Drive", so I oppose "acd" as an abbreviation. ad, adrive or amazondrive work though. Which one would be most preferable? I also believe the acdcli backend should be removed, so "compatibility" to that backend's name is not a good argument for me. See also point 4 on that.

2.
I'm confused. I documented this, and it's visible in the diff on launchpad. Did you maybe miss that I pushed two commits to the branch?

3.
If anything, the old backend should be updated to reflect the new name. The product has been renamed to just "Amazon Drive" months ago. 

4.
Yes, Kenneth didn't document it when he merged that backend originally. I don't see the point in documenting it now, given
- there will be a native backend that can be debugged more easily (full stacktraces, no version incompatibilities)
- it doesn't bail if there is no TTY, hanging forever when it needs to oauth
- concurrent usage of acdcli can break its database storage, which pops up in the acdcli bugtracker *a lot*. To be fair, this doesn't happen with duplicity often because it runs as root, while most other acdcli operations are run with a user account where a separate DB is maintained. Still, this kind of issue cannot happen with the native backend at all.
- it calls the "get_changes" endpoint rather often, which is throttled with acdcli being on developer limits again. Long story short, acdcli does more work than needed for the way duplicity works, possibly slowing down/breaking the backup depending on how the Amazon Drive account is used otherwise.

I decided not to submit the acdcli backend specifically because it didn't meet my quality expectations, so please understand I would rather deprecate it than spend time on documenting or supporting it any further. 

Cheers,
Stefan
-- 
https://code.launchpad.net/~breunigs/duplicity/amazondrive/+merge/309084
Your team duplicity-team is subscribed to branch lp:duplicity.