I currently scan the Dagger.jl and DTables.jl documentation and one thing which came to my attention is that while a DArray supports broadcasting, a DTable does not.
I think it would be great for users if a syntax like
mycustomfunc.(mydt.anothercol) would create a DArray, ormydt.mycol .= mycustomfunc.(mydt.anothercol) could update the DTable itself, or if this is not possible because of no-mutation policymynewdt = mydt.mycol .= mycustomfunc.(mydt.anothercol) could be an alternative syntax.
My motivation is to take mental burden from the user if they like broadcasting, so that they don't have to decide between DArray or DTable, but they just work nicely together like Arrays and DataFrames do so already.
I currently scan the Dagger.jl and DTables.jl documentation and one thing which came to my attention is that while a DArray supports broadcasting, a DTable does not.
I think it would be great for users if a syntax like
mycustomfunc.(mydt.anothercol)would create aDArray, ormydt.mycol .= mycustomfunc.(mydt.anothercol)could update the DTable itself, or if this is not possible because of no-mutation policymynewdt = mydt.mycol .= mycustomfunc.(mydt.anothercol)could be an alternative syntax.My motivation is to take mental burden from the user if they like broadcasting, so that they don't have to decide between DArray or DTable, but they just work nicely together like Arrays and DataFrames do so already.