Brodie Gaslam

Could be confusing. Also maybe should check that numerics are in integer range and not just rounded. Docs for `?vet` and `?alike` should all be clearer about this, perhaps with...

``` vetr::vet(base::sum(), 1) ## Error in vetr::vet(base::sum(), 1) : ## Internal error: no integer codes produced by parsing process, which should not happen; contact maintainer. vetr::vet(sum(), 1) ## [1] TRUE...

See https://github.com/brodieG/fansi/issues/76. To reproduce copy tests directory out of original package folder structure and try to run tests from there. h/t @tillea for identifying and reporting problem.

The problem is that https://github.com/brodieG/unitizer/blob/5ff79c3103d81646b3c8ce971124519fc4133d81/R/heal.R#L157 causes the assign form of an active binding to be called, and if the active binding does not accept the extra parameter for setting things...

Maybe percent of lines or functions. One computationally expensive but maybe close to doable way to do this is to compute the coverage of the dependency when running the tests...

It fails, and then it also leaves `stdout` in a broken state. ``` > library(dict) > d entries(d) $ color : Error in print.default(value, digits = digits) : promise already...

I'm guessing this is happening due to a combination of documented difficulties in building R with the GCC UB sanitizer, and the problems caused by ~/.R/Makevars affecting all builds, but...

Based off of [@wrathematics observation](https://twitter.com/wrathematics/status/921016665529356289): ``` > library(inline) > T T [1] TRUE > T == TRUE [1] FALSE ```

Should be possible to find a different environment that breaks things in the same way but doesn't trigger the parent env loop check. https://github.com/brodieG/unitizer/commit/26dc83293eaf3dc213f97394c64a3389575c56cc

this isn't great: ``` vet(NUM.1, Inf) [1] "`Inf` should contain only finite values, but does not" ```