This book page contains an overview of all desired functionality for the first major rewrite of Hierarchical Select, which will be version 4.

After it has been decided what the scope will be, this will be transformed into a roadmap.

All issues mentioned below can be found in one handy overview by browsing the HS issue queue by the "HS4" tag.

Back in February 2011, I was anticipating to do an internship (that would complete my studies) at a Drupal shop. I did it for Facebook instead. If it were at a Drupal shop, I would probably have worked on client work, but also on the HS4 rewrite. Alas that did not happen.
For now, it is not clear when it will happen after all. The rewrite is postponed indefinitely. In a few weeks, I'll have decided which company to work for. Once that has happened, I will be able to say decisively whether the rewrite will happen or not. It greatly depends on the company I'll be working for.
I apologize for any disappointment and reliance on the ETA I had provided and failed to update, even though it said "at the earliest".

The following will definitely be implemented:

• Drupal 7 only
• Unit tests: #397848: Tests for Hierarchical Select!.
• Drastically improve performance
  • #544324: Performance: Improve multiple item processing in big taxonomy hierarchy setups. (might still be committed to D6/HS3)
  • #528156: Performance: Remove Hierarchical Select's form cache and don't reconstruct the entire form anymore
  • #492068: Change HS API to support lineage storage provided by implementations
  • #790234: Performance: hook_hierarchical_select_root_level is called twice by _hierarchical_select_hierarchy_generate
  • #578984: Refactor _json to use the cache rather to reconstruct the whole form on every ajax request.
  • #448316: New hooks hook_hierarchical_select_offspring_count() and hook_hierarchical_select_children_count()
• Develop a proper API for the client-side caching system, ship with two implementations: HTML5 client-side databases (already exists in HS 3) and JS
  • #370505: Performance: revamp HS' cache system
  • #284638: New cache system: JS
• Ability to replace the HTML "select" elements of HS with other elements, such as autocompletes. Also make it possible to change the behavior, e.g. no incremental level display, but always showing all levels, or disabling animations (or replacing them with other animations).
  • #920244: Allow autocomplete option for one of the levels
  • #654750: Statically display all levels: no animations, no incremental level display
  • #795242: Large taxonomy term names in several lines
  • #351872: Per-level autocomplete fields
  • #787988: Different level labels depending on taxonomy parent term
  • #442546: Ability to select multiple items in the deepest level, to add them all at once to the dropbox
• Support for more advanced use cases
  • #1048344: HS exposed Views Filter not respecting Views Relationship
  • #312400: Allow the selections in the dropbox to influence the hierarchy
  • #298611: Granular role-based permissions
  • #679168: Custom HS implementations: through UI instead of code by (ab)using Views!
  • #539156: Optional hook: hook_hierarchical_select_item_link()
  • #326300: Adding extra information to the page after each selection using Javascript
• HS UX improvements:
  • #321252: Layout setting
• Admin UX improvements:
  • #253950: Live preview of Hierarchical Select
  • #321257: Make yes/no options checkboxes, not two radios

Blissful notes:

• No more need for taxonomy_override_selector in Drupal 7. This is what caused the zillion compatibility issues with other Taxonomy form overriding modules. Bliss!
• No more need to support the extremely poor Content Taxonomy module in Drupal 7, which has more bugs and compatibility issues that any sane person could imagine. Wonderful!
• No more "Invalid response from server" because of missing PHP includes due to badly written modules, i.e. no more need for $form_state['form_load_files']. This problem has triggered hundreds of support requests in the D6 version of HS. Yay!

These blissful changes will significantly reduce the number of support requests I'll have to handle, therefor allowing me to focus on HS itself much more.

(This page is continuously being updated.)