HTML5 and figure element

it would appear that this is all set in CKEDITOR.dtd.$block. i also discovered that newer versions include the HTML5 elements and at first glance have them defined the way i want them.

so, time to upgrade!

on another note, is there any reason why it wouldn't be possible to dynamically change what is a block element?

the upgrade does in fact solve one of my problems. when the figure element is inserted into the editor, it does not have a surrounding paragraph element. however, once i save data and reload it, my img and the span tags inside of my figcaption tag have surrounding p tags.

i am aware of the solution of change config.entermode, but this does not work for me. in general, i do want a space to translate to a paragraph tag, but not if the element in question is already surrounded by a block element. when i was looking at CKEDITOR.dtd, some of the comments seemed to suggest that the editor will take into account what an element is surrounded with. does that mean i am experiencing a bug?

after further analysis of dtd.js, i realize that figcaption is not included. i would guess it should be. i have yet to repackage ckeditor, so i am going to work on that and do some tests.

as for the img inside of figure. i believe it is acting well except in cases where a specific plugin is acting on the img. as in alignment. if i try to align the image, it's only choice is to wrap it in p tags and apply the class to that. i guess this is correct. perhaps i need to find a way to make the figure selectable, instead of the image, so that alignment can be applied there.

if anyone has any experience with this, i would appreciate the help.

ok, i managed to get figcaption into the dtd.js and recompile it. not sure where i can submit this patch or whether anyone cares.

in the end, my issue may go deeper into how ckeditor works. i would love if anyone can comment on this behaviour...

i have the following markup:

<figure class="full">
<img alt="logo" src="images/logo.png" />
<figcaption>
<span class="caption">Cool Logo!</span><span class="credit">by Me</span>
</figcaption>
</figure>

if i click on the caption or credit in the editor, both of my spans inside of the figcaption are surrounded by a p tag. is there anyway to avoid this? in my mind, figcaption is a block level element (since i've now defined it so). i would imagine that as long as i am click and editing the contents of the span, i shouldn't need to add p tags. i suppose if i click outside of the spans, then a p tag should be created.
the same is true for the image. if i click to the side of the img, p tags are added around the img. i suppose i can understand why this works this way, but i have two questions:

1) if no text is added, i don't think the p tags should remain around the image.
2) there needs to be some logic that accounts for the fact that my img is surrounded by a figure.

i would love to hear from one of the developers on this.