We would like to find a first small slice of the application to implement that can flex our tech stack end to end, interact with important systems, and prove out some foundational modeling.
A good candidate might involve a feature that is based on the relationship between Veterans and representatives, like showing the Veterans associated to a representative. This might prove out some of the modeling around roles and authorization. It would build on top of our authentication approach as well.
This epic is about making the determination of what to build rather than any of the implementation.
Alex
Yep. In my mind, we could have one engineer (ideally someone with some experience with the codebase) take the lead and then have a larger session/sessions with everyone to share out learnings and everyone can trace through the codebase together
Alex
Had a chat with Bill Chapman, the OCTO director of engineering, yesterday. He offered two suggestions for the ARF team:
We should think through an abstraction layer for rep-facing tools that sits between the frontend and any vets-api code paths we want to use. This layer should probably handle:
whitelisting actions that rep-facing tools can take
auditing and monitoring of actions that reps are taking in production
any identity “transformation” to ensure we can use vets-api functionality with the rep as the user rather than the Veteran
Vets-website is already a well-compartmentalized codebase and folks don’t have concerns with building the rep-facing tools frontend in that repo — but when we get to the point where we’re repurposing existing forms, functions, etc, we need to be aware that any changes we make for rep users create the risk of regressions for Veteran users. We’ll likely need to develop some processes/conventions for any rep-facing teams to coordinate with the benefits portfolio teams that own/maintain the original forms.
Needs refinement
We would like to find a first small slice of the application to implement that can flex our tech stack end to end, interact with important systems, and prove out some foundational modeling.
A good candidate might involve a feature that is based on the relationship between Veterans and representatives, like showing the Veterans associated to a representative. This might prove out some of the modeling around roles and authorization. It would build on top of our authentication approach as well.
This epic is about making the determination of what to build rather than any of the implementation.
Context Some discussion with Alex
Some more discussion with Alex