If there is really just the one decision point then any option seems fine and within Elixir’s idiomatic style, including the way you’ve done it.
The with syntax would make your code something like:
with socket <- assign(socket, slide_navigator_menu_state: put_in(slide_navigator_menu_state.folded, fold?),
maybe_reset <- Tools.maybe_do(!fold?, &reset_slide_management/1) do
fold_slide_menu(maybe_reset, !fold?)
else
_ -> fold_slide_menu(socket, !fold?)
end
This is clearly not an improvement with the style and structure and context you’re actually using. But if you have multiple transforms that could fail or multiple conditional branch points you might derive more benefit from the with approach.
with {:ok, new_socket} <- imaginary_failable_assign(socket, slide_navigator_menu_state: put_in(slide_navigator_menu_state.folded, fold?),
{:reset_slide, reset_slide} <- maybe_reset_slide_management(new_socket, fold?),
{:reset_block, reset_block} <- maybe_reset_block_management(reset_slide, fold?) do
fold_slide_menu(reset_block, !fold?)
else
{:assigns_err, assigns_err} -> # log the error and return the original socket or whatever you like
{:slide_not_reset, not_reset} -> fold_slide_menu(not_reset, !fold?)
{:block_not_reset, not_reset} -> fold_slide_menu(not_reset, !fold?)
end
Again, with this particular case I don’t think that’s any real improvement, but if you like that granular control over logging messages or handling various branchpoints in one flow it can be nice.
