ext/afform/docs/roadmap.md

   1 # Roadmap
   2
   3 The `afform` extension is a proof-of-concept.  It aims to demonstrate the core model/concept in which AngularJS
   4 provides the standard data-format and component-model shared by developers (working in code files) and administrators
   5 (working in a programmatic GUI).
   6
   7 As a proof-of-concept, it is necessarily incomplete.  In particular: (a) some functionality is envisioned for
   8 additional extensions and (b) within this extension, there are known issues.
   9
  10 ## Suite of Extensions
  11
  12 This extension is expected to be the base for a suite of related extensions:
  13
  14 * `afform`: Base framework and runtime. Provides APIs for developers.
  15 * `afform_html`: Present web-based editor for customizing forms in HTML notation. Use Monaco or ACE.
  16   (In the near term, it's a simple demonstration that forms can be edited by users in a browser; in the
  17   long term, it occupies a smaller niche as a developer/power-admin tool.)
  18 * `afform_gui`: Present a web-based editor for customizing forms with drag-drop/multi-pane UX.
  19   (In the long term, this is the main UI for admins.)
  20 * `afform_auditor`: Report on upgrade compatibility. Test changesets. Highlight dangerous/unknown/unsupported form elements.
  21    Score maintainability of the system.
  22
  23 ## Known Issues
  24
  25 Within this extension, there are things which need updating/addressing:
  26
  27 * There are several `FIXME`/`TODO` declarations in the code for checking pre-conditions, reporting errors, handling edge-cases, etc.
  28 * Although afforms can be used in AngularJS, they don't fully support tooling like `cv ang:html:list`
  29   and `hook_civicrm_alterAngular` changesets. We'll need a core patch to allow that. (Ex: Define partials via callback.)
  30 * We generally need to provide more services for managing/accessing data (e.g. `crm-api3`).
  31 * We need a formal way to enumerate the library of available tags/directives/attributes. This, in turn, will drive the
  32   drag-drop UI and any validation/auditing.
  33 * Haven't decided if we should support a `client_route` property (i.e. defining a skeletal controller and route for any form).
  34   On the plus side, make it easier to add items to the `civicrm/a` base-page. On the flipside, we don't currently have
  35   a strong use-case, and developers can get the same effect with `civix generate:angular-page` and embedding `<div hello-world/>`.
  36 * Injecting an afform onto an existing Civi page is currently as difficult as injecting any other AngularJS widget --
  37   which is to say that (a) it's fine for a Civi-Angular page and (b) it's lousy on a non-Angular page.
  38 * The data-storage of user-edited forms supports primitive branching and no merging or rebasing.  In an ideal world
  39   (OHUPA-4), we'd incorporate a merge or rebase mechanism (and provide the diff/export on web+cli).  To reduce unnecessary
  40   merge-conflicts and allow structured UI for bona-fide merge-conflicts, the diff/merge should be based on HTML elements and
  41   IDs (rather than lines-of-text).
  42 * API Request Batching -- If a page makes multiple API calls at the same time, they fire as separate HTTP requests. This concern is somewhat
  43   mitigated by HTTP/2, but not really -- because each subrequest requires a separate CMS+CRM bootstrap. Instead, the JS API adapter should
  44   support batching (i.e. all API calls issued within a 5ms window are sent as a batch).
  45 * Default CSS: There's no mechanism for defining adhoc CSS. This is arguably a feature, though, because the CSS classes
  46   should be validated (to ensure theme interoperability).