Spike: Evaluate Riff Plugin Development for Mattermost 5.16

[adonahue]

As the CEO, I want to understand viability of internally developing our own Mattermost plugins, so that we can make a decision about the plan and the timing to upgrade our EDU product to the current version of Mattermost.

The goal of this spike is to answer the following questions:

• How hard / easy is it to develop plugins internally with the developers we have?
• How much of the Riff code base (e.g. LTI) is not separable (as a plugin) from our Mm implementation and will require us to continue to handle it as a manual merging process when migrating to new Mm versions?
• What are the estimates for us moving the Riff code to plugins (dashboard and video)?
• Are there any other factors that would block us from starting the plugin, or otherwise have negative impacts (e.g. we can't build plugins until the packaging is complete)?

Spike Acceptance Criteria

• [ ] Limit the spike to three days of dev time; if that is not enough time, let Amy know.
• [ ] Do an analysis and summary of the Riff code base to understand what code is separable and could be shunted into a plugin. The summary should include:
• [ ] an estimate of how the code is divided (as a percentage of the code base) e.g. 10% inseparable / 90% separable;
• [ ] a description of each part of inseparable code (e.g. LTI code);
• [ ] a rough estimate (hard, med, easy) of how difficult that would be to handle in a manual merge process to with future versions of Mattermost

Proof of concept plugin Create a proof of concept for the video player in Mattermost 5.16 which has the ability to:

• [ ] Launch the Riff video player from Mm, as a plugin.
• [ ] Have a working video call.
• [ ] Create a Riff Dashboard container as a Mm plugin.
• [ ] Talk to Riff server, get a piece of data (any data), and display it in the dashboard.
• [ ] Talk to Mattermost server, get a piece of data (any data), and display it in the dashboard.

[adonahue]

Per @jordanreedie - this is persuble, but it requires a DB modal that gets invoked from a plugin button. The rest of the prototype work is reasonable to do, but still needs to be done.