jdanish
commented
4 weeks ago The basic organization you note sounds right. A minor note that since the sentence starters are not targeted, we've largely been just using "(see notes above)" or something like that. It's fine to leave it as-is, or long-term we could just remove that since we aren't doing much with it.
Adding a new resource will probably require some kind of FTP/S3 interface. So adding files like PDFs will require some technical chops.
Assuming you can still have evidence point to a url, this is totally fine since the teachers can always upload google docs to their own account and link to them or similar. Long-long term we might want to build out or leverage an interface to make ftp not necessary, but short-term if the teachers who don't know how can simply use external links, all good.
Adding a resource after the study has already begin should not be a problem. Though deleting them might mess up the resource IDs.
That's fine as we can just hide mid-study. Though we want to gracefully handle hidden resources (e.g., either hide their e evidence links, or show a popup that says the teacher hid it so ask them).
We assume that for a given location, they would be fine sharing the resource library -- e.g. you're not dealing with two completely different domains where sharing the resource IDs will be a problem. e.g. you won't have 100 different resource items and turning them on and off for different sets of classrooms is not a challenge.
Agreed. If we really need them to be different we can just run a separate server, right? Like if there was a bunch of 6th grade teachers doing one thing, and 7th grade teachers doing something completely different but in the same building, we might opt to use 2 servers for that. My hope is that AWS will be easy enough that it is far easier to run 2 servers then to solve all of the UI problems required to have one server server 2 teachers who are doing entirely different things. The sharing is for teachers doing overlapping things, which happens quite often.
The one other / related issue I'll note is around the resource ids. Right now, there is no easy way to re-order them, which means if we use a server, test some ideas, move things around, etc., then a given class might end up with ids 8-15. That feels a bit confusing. Ideally we'd want to have some control over that. It's OK if they are the same across classes on the server, I think, but if we could at least re-order them that'd help make it feel intuitive. Ideally, this would be doable even after the study starts, so I imagine you'd want to have a separate unique ID for each resource, and a display ID, so that we can change the display ID without losing the actual resource. We will avoid doing a lot of changing mid-study because kids might remember that 3 was fish, and 4 was algae, but in the cases where we made 5 be predators, have not yet made it visible, and instead move that to 6 so that we can add a new 5 that is the toxins kids assumed are there ... it makes more sense to be able to keep those in order.
benloh
Clarifying the number of servers per classroom/site.
@jdanish We want to make sure that our current implementation architecture works for what you need for the studies.
We had talked about having two "servers": one for each location.
Each "server" install has a shared set of resources.
One key question is how Resources are handled. Currently Resources IDs are shared across all classrooms on one "server" -- so the assumption is that the resource IDs (e.g. the "footnote" numbers for the resource library) are persistent across all classrooms on the server. There are a few implications:
The AWS architecture will require a different mapping of database files and resource library items and may be shared across "servers" so we want to make sure that the capabilities and limitations of that structure make sense.