[WikiEN-l] Why we need a good WYSIWYG editor

[William Pietri]

On 01/04/2010 11:41 AM, Gregory Maxwell wrote:
> Bad presentation in the edit isn't, in my view, the biggest problem
> with WYSIWYG systems the problem is that they frequently behave
> inscrutably, even ones designed from the start as WYSIWYG (as opposed
> to boltons as we'd have).  Issues like... "Help! in order to un-intent
> this I have to copy, delete, paste and reformat!" or "I pasted this
> bit and everything turned bold or vanished and now I can only fix it
> by throwing out all of my edits!"
>    

Maybe this is drifting too much into the technical details, but I 
suspect that  is a symptom of popular approaches to developing WYSIWYG 
tools, rather than a necessary fact of WYSIWYG tools.

The number of combinations of actions a reasonable user can get up to in 
a serious editing session is just too high for a developer to track 
mentally. So WYSIWYG systems tend to reach [[Critical Mass (software)]] 
quickly, and obscure bugs become endemic.

I expect that we could do better with:

   1. Research into popular existing strains of editor to see pros and
      cons of various approaches,
   2.  From that, a clear theoretical model,
   3. Extensive automated unit tests, run regularly in all supported
      browsers,
   4. Instrumentation of the editor to track user behavior and
      automatically report errors, and
   5. A rule that no bug gets fixed without automated tests that expose
      the bug.


I think we could stave off critical mass and keep painful errors pretty 
low with an approach like that.

William