Suggestion: Register the package

[briochemc]

I find myself regularly coming back to this issue that made you build this package in the first place (combining some autodiff and Unitful). Maybe registering it will bring more attention to it and help develop it further?

[AleMorales]

My main issue is that I do not have any time at the moment to maintain this package and I am actually rethinking the whole idea of combining Dual Numbers with physical quantities, so I do not want to commit to anything long term (and in my view, registering implies such commitment). There is also some performance issues (type inference fails sometimes with, e.g. power functions) that would need to be addressed before I can recommend this package. In any case, why do you think that registration will bring more attention?

[briochemc]

OK no worries!

I'm not sure it would bring any attention, but I just thought it might, simply because it would provide access from the REPL :)

[AleMorales]

OK, but Pkg supports working with unregistered packages just by using the URL of the git repo instead of the package name. This is described in the official documentation but I can add it to the README if that helps.

[briochemc]

Oh I don't think you need to add anything. I just meant that the REPL would "know" about it in the sense that autocomplete from (v1.2) pkg> add Unit shows registered packages only:

(v1.2) pkg> add Unit
UnitfulAngles      UnitfulCurrency     UnitfulUS           UnitfulMR           UnitlessFlatten
UnitfulAtomic      Unitful             UnitfulAstro        UnitfulRecipes      UnitfulIntegration

But again no worries, especially if you have plans for

rethinking the whole idea of combining Dual Numbers with physical quantities

which sounds exciting IMHO!