Pale Moon 25 will employ the alternative implementation.
Since there are continued problems with the sync service, an "other" project that has been tossed into my lap to take care of because of Mozilla's move to an incompatible "Firefox Accounts" and "Sync 1.5" setup (they should have made it 2.0), a few points in this post to keep you all informed about what's going on:
- Pale Moon Sync was set up in accordance with the "run your own sync 1.1 server" documentation available at Mozilla. This involves:
- A complex python implementation of Weave that does not exactly act like a normal process (service rerstarts fail, for example, and the "worker threads" respond very sluggishly). It also doesn't seem to be very scalable.
- Using egg/gunicorn as an interface to the front-end, which has already caused issues of its own
- Passing a whole bunch of things through NGinX to gunicorn, which may be at the root of the authentication issues encountered by people
- Additionally, the setup causes load issues when not run on a dedicated server with lots of throughput and iops available.
- Because of 1&2, I've been forced to find a different solution. I have found one implemented in PHP, but it's a little tricky to set up. Also, this is not a "full" server implementation yet so some features may be missing as a result, like password recovery options. I'll have to investigate the extent of implementation available.
- It looks like (3) can be made to work properly, and I'll use a new server with SSD caching to hopefully avoid (2) altogether, even on a more efficient and less fragile implementation, since there will still be a lot of database reads/writes.
- Sync will be down while this is being set up. The current implementation is obviously broken, and I'd rather avoid needless support requests.
- Keep in mind that sync is an additional convenience service for Pale Moon provided at no cost to users. I do not have an SLA or even a requirement to keep it available, but wish to do so, anyway, to provide a better experience for Pale Moon users. If maintenance effort or expenses for the sync service get out of hand, I will have to discontinue the service, but I will do my best to keep it available for everyone free of charge.
- I may want to ask for help from or hire an experienced PHP programmer to expand on the implementation in PHP (FSyncMS) to add any missing features there may be for account management.