Júlio Hoffimann

I've started looking more carefully into the changes and realized that we already define `isapprox` as part of the public API. I believe we just need to specialize the methods...

> You can avoid such non-reproducible failures by putting `Random.seed!(42)` at the beginning of your `runtests.jl` It is in the plans. Just trying to sort out this isapprox definition first....

@ettersi can we port some of these changes to the latest main branch?

Thank you @ettersi ! Closing this in favor of #294

> Probably we should load anything and let the user fix it afterwards. But we could have a keyword to check for validity and add a warning? Maybe adding the...

We added the code and corresponding type as discussed in https://discourse.julialang.org/t/esri-code-for-british-national-grid-not-known-by-geoio/117641 I will leave this issue open because we still need to work on a final missing feature to make...

It looks great! Looking forward to the tests. Please let us know when they are in place. The decision to specialize the backward formula is sometimes made based on comparisons...

The error message in the tests suggests that the formulas are not preserving the floating point type: ```julia MethodError: no method matching allapprox(::GeodeticLatLon{NAD27, Quantity{Float64, NoDims, Unitful.FreeUnits{(°,), NoDims, nothing}}}, ::GeodeticLatLon{NAD27, Quantity{Float32,...

Regarding the comparison with PROJ, I can try to find some time to read into the docs to see if I can spot some issue with the translation from the...

> Proj was outputting feet instead of meters. I am glad we decided to be explicit about units 🙏🏽 Some tests are failing with `DomainError`: ```julia LatLon LambertConic: Error During...