A couple points and a question; please bear with the scattershot:
There appears to be a point of view, where the cabal thing is a stagnant piece
of technology that only serves to split the community and confuse newcomers
by detracting them from the easier, better way.
In such a view, it wouldn't be too big a leap to come to an understanding,
that phasing out cabal from all avenues does a good service to the
community.
If anyone does indeed have this view, please be assured that it is far from universal. Cabal is essential for many Haskell workflows. I and other practitioners use Stack for some projects and Cabal for others because both tools have their strengths, weaknesses, and mutually unique capabilities. The view that Cabal is stagnant is very outdated, most notably with respect to the overall pace and vigor of the Cabal issue tracker, and the excitement around the `cabal new-build` Nix-style commands.
-1. The topmost method will target beginners of the language and
should be Stack even if HP comes with Stack.
Many beginners will be picking up a copy of a book, or working through an established curriculum. While the cutting edge of published guides and course curricula may have converted to Stack, there is a vast quantity of material for beginners that assumes you can type `ghci` and `cabal install` at the prompt. Which particular materials to recommend to a beginner is a matter of reasonable debate, but the best support we can give to those beginners as stewards of the infrastructure is to highlight an option that allows beginners to quickly succeed with the option of their choice.
should be Stack even if HP comes with Stack.
What is the reasoning here? How does having other tools bundled take away from the new user experience?