FLUID-4132

[Antranig Basman]

Hi Cindy - I tested your branch just now and it doesn't pass the test I mentioned. If you navigate to
src/webapp/demos/inlineEdit/rich/demo.html , operate the CKEditor 3 instance of the rich widget, and then 
save it, you will see "Undo edit" appear. In fact the original regexp doesn't pass this test either, 
suggesting that this issue was not checked when the FCKEditor integration was upgraded to CKEditor. Could 
you look into what change is caused by CKEditor during this roundtrip and adjust your regexp so that it 
tolerates it?

Thanks,
Antranig.


PS - the issue appears to be caused by the mysterious appearance (and then disappearance) of an empty attribute

title=""

on some randomly chosen element in the rich text material, either on editing, or on saving, which actually 
afflicts both of the controls - I thought I saw it work once, but TinyMCE seems to fail consistently for me 
too. I am testing on FF 3.6 on Windows. I don't remember seeing anything like this on the original demo and 
so am wondering whether this is something caused by some of the other JS on the page, e.g. chili, since it 
is somewhat mysterious that it affects both of the widgets in the same way. Please look into this :)

Here are some sampled values to get you started:

http://pastebin.com/rRmsNW7L



On 02/06/2011 13:11, Antranig Basman wrote:
> Hi Cindy - the purpose of the regex is actually to canonicalise HTML tag names rather than user text, so
> your test should include editing a field which contains an initial value which is some markup. THe issue is
> not so much that of "loss" as of false detection of change - so the acid test is to ensure that a field with
> an ORIGINAL rich text value can be edited, and immediately saved without triggering change detection. The
> tests you have performed are useful too - but those can actually be done in automated tests rather than
> needing to be done manually.
>
> Cheers,
> Antranig
>
> On 02/06/2011 13:02, Li, Cindy wrote:
>> Hi Antranig,
>>
>> Following up yesterday's dev meeting, I'm helping with testing the regex that was modified in this jira.
>>
>> What I've done:
>>
>> 1. I came up with a string that contains all the keyboard characters and letters in lower& upper case
>> 2. input the string into inline edit fields of these demos:
>>
>> src/webapp/demos/inlineEdit/simple/html/inlineEdit.html
>> src/webapp/demos/inlineEdit/rich/html/inlineEdit.html
>>
>> 3. ensure none of the characters in the string was lost during edit, save, undo, redo.
>>
>> So far so good.
>>
>> Do you think this test is decent enough? Anything else I should test? Thanks.
>>