Temas touched on this a bit in his last blog post, but I wanted to write a more detailed explanation of where things are at for synclets. The short synopsis of synclets is that they are a simple way to pull data from a provider and feed it into the system. Synclets are basic routines that are fed authentication keys and some configuration and uses that information to pull down data from the provider which is funneled back into locker core via JSON. They are a drastic reduction in scope for what’s currently required for a connector. Adding new connectors currently relies on the developer managing a lot of pieces (authentication, running a web server, processing data from the source, feeding that data into Mongo, generating correct events) and since the majority of those pieces are fairly common across all of the connectors, we want to drive towards providing a system that manages these common components.
The first step towards this end was to convert the stable connectors into packages composed of synclets being managed by the common code. This has been completed (we’re still working on cleaning up some of the patterns to get them right, but we have a start checked in that we’re playing with) and they are now ready for people to start poking at. The current plan is to eventually migrate all of the connectors (and all the data your connectors have been collecting!) to synclet powered versions. For the time being, existing connectors won’t be affected by any of the work being done around synclets.
The missing pieces that will tie this work together and make it easy for developers to implement their own synclets mostly surround authentication and creating a UI for managing installed synclets. We will be implementing something like everyauth to manage authentication for synclets at some point in the near future. This will enable us to simplify both the UI, and the implementation, for authentication and allow us to provide authentication keys to any number of synclets that would like to pull data from a provider.
What does this mean to developers hacking on connectors today?
Not a whole lot just yet. We want to spend some more time ensuring that we have this pattern right, and connectors will continue to exist exactly as they are today during that period. If you’re feeling super adventurous, feel free to poke through some of the code in lsyncmanager, and look at the new synclets that are now provided with the connector code for Facebook, Twitter, Github, Foursquare, and Google Contacts. Once the authentication and UI pieces have been baked into the project, we’ll write a more detailed post describing exactly how you would use those pieces to build synclets.