Closed bobular closed 1 year ago
All of the above seems correct, but focused on continuous variables. We also need to remove from the categorical axes ticks any values explicitly filtered against.
All of the above seems correct, but focused on continuous variables. We also need to remove from the categorical axes ticks any values explicitly filtered against.
This should be happening now, with "filter aware" constraints.
Just FYI, I have tested this categorical axis stuff with Lineplot at qa site and it seems to work properly. Here is an example with Age group at enrollment as an X-axis variable and subsetting 12-23 and 24-59 months (0-11 months are not selected). The tick concerning 0-11 months is not shown accordingly.
Current behaviour for axis ranges in "Full" mode
start: minimum of
end: same as above but with max and no zero
Desired behaviour for axis ranges in "Full" mode when NOT filtering on this axis variable
same as above
Desired behaviour for axis ranges in "Full" mode when filtering on this axis variable
start: minimum of
end: maximum of
So as you can see, it won't necessarily affect variables with hand-annotated displayRanges. That is OK. @d-callan confirms!
For continuous overlay variables, the legend range should use the same as above except for no special zero handling.
Note also that the custom and auto zoom axis modes are unchanged by this issue. Also note that there are no plans for auto zoom or custom ranges for overlay variables. They are always in "Full" mode.
Implementation-wise, I can imagine passing
filters
to theuseDefaultAxisRange
hook and letting that do all the work. The study variable metadata is NOT changed when adding a filter for that variable.