Tests are failing but I'm somewhat suspecting it's not actually my fault. It seems that the change in 7dcfca5fdb3046f6d13e3c32824f0ba0a7d08038 reveals a hidden bug. The age.group variable in contact_matrix() is actually never defined. data.table was quite happy with it being NULL, as set initially but now it's registered as an actual global variable, it doesn't work:
This hypothesis is confirmed by the fact that tests pass fine when I revert 7dcfca5fdb3046f6d13e3c32824f0ba0a7d08038, even though it should be an entirely benign change.
This PR may seem quite overwhelming if you look at it as a single change but commits are separated in quite clearly distinct changes so it should be much easier if you review commit by commit.
sbfnk
commented
2 years ago
Tests are failing but I'm somewhat suspecting it's not actually my fault. It seems that the change in 7dcfca5fdb3046f6d13e3c32824f0ba0a7d08038 reveals a hidden bug. The
age.groupvariable incontact_matrix()is actually never defined. data.table was quite happy with it beingNULL, as set initially but now it's registered as an actual global variable, it doesn't work:https://github.com/epiforecasts/socialmixr/blob/c83358240dc751602ec4fb30c5dc901590563707/R/contact_matrix.r#L550-L555
This hypothesis is confirmed by the fact that tests pass fine when I revert 7dcfca5fdb3046f6d13e3c32824f0ba0a7d08038, even though it should be an entirely benign change.
This PR may seem quite overwhelming if you look at it as a single change but commits are separated in quite clearly distinct changes so it should be much easier if you review commit by commit.