Store different report types in different collections

[dgorissen]

in particular: currently waterpoints are stored as reports, this does not make sense as they are not reports, belong in separate db, outside of core taarifa

[nicowe]

What exactly are we trying to achieve? @Dirk: do you mean a seperate database or a seperate mongo collection?

Currently Waterpoints are a subclass of the generic baseclass report in the ODM. Do you want to change this?

[dgorissen]

Its just subtle semantics. See my (crappy) picture I emailed a couple days ago about the high level picture.

Right now submitting a new report is equivalent to submitting a new waterpoint (you enter location, ward name, technology, village, etc). While actually a report about about a waterpoint should contain, for example, waterpoint id, who reported by, date, problem recorded, comments, ...

So ideally a Taarifa report just refers to a particular waterpoint (by id) as stored in openstreetmap (where the waterpoint data should ideally end up) and does not duplicate waterpoint attributes in the reports collection. Of course for now we have to store the waterpoint CSV data somewhere. We can keep it in the same db but I would store it in a separate collection to keep the semantic distinction.

[dgorissen]

Just to detail, given an empty, raw, database the upload script should:

• create a waterpoints collection and upload the waterpoint csv data into it. See if a sensible _id field can be generated. This collection is essentially read-only and not changed.
• create the usual reports/services collections and create a couple of sample/dummy reports. Each report refers to a particular waterpoint id with fields such as: timestamp, waterpoint id, reported by, status, comment, ... The service/report schema definitions will thus have to be updated. Care must be taken to ensure referential integrity (this is where mongo worries me).
• update the UI so it pulls data from the waterpoints collection and reports collection, joined on waterpoint id