This simplifies, though not as much as I was hoping, the patterns for accessing data along strict type patterns. I may see better results once I'm getting the Time/Distance views working.
Step and weight records may be presented in any order. Any test that
tries to enforce that one gets presented before teh other can't cannot
succeed. So, I've removed that test and instead put in a warning that
will appear when the view model gets loaded.
DayDetailViewModel needs testing. I've worked out an improved API, and a set of tests to go along with it, and those can be made more easily with a mockable RecordProvider. So, in addition to stubbing out a bunch of tests, I've also created RecordProvider, mocked it, and implemented it for App.
This is preparatory work. Having the view model directly retrieve data both adds a degree of symmetry (it both gets data from and sends data to the app) and makes it possible for the view model to refresh itself when needing to revert data or after saving data.
I've created the view model and added a getter function for the weight.
I'm passing the view model now to the DayDetailView, DayDetail, and
DayEdit.
I'm starting to set up the Save function for the view model, draining
all of the updated records and saving them.
None of the components yet save any updates to the view model, so
updated_records is always going to be empty until I figure that out.
DayDetail, the component, I used to use as a view. Now I'm swapping
things out so that DayDetailView handles the view itself. The DayDetail
component will still show the details of the day, but I'll create a
DayEditComponent which is dedicated to showing the edit interface for
everything in a day.
The swapping will now happen in DayDetailView, not in DayDetail or an
even deeper component.
This is a super tiny data structure that covers an edit mode, a view
mode, and an unconfigured mode. It's mostly a container so that views
don't have to preserve everything directly.
I move the weight edit view into the validated text entry widget, and I
work on some of the unfortunate logic in the weight blur function. I've
left behind a lot of breadcrumbs for things that still need to be done.
Savanni D'Gerinel