Am 13.03.2011 21:46 schrieb Aryeh Gregor:
2) How much work should we go to to produce nice-looking markup?
E.g., if the user unbolds "baz" in
<div style="font-weight:bold">
<p>Foo
<p>Bar baz
</div>
should we produce something like
<div>
<p style="font-weight: bold">Foo
<p><b>Bar</b>baz
</div>
like WebKit does, or would it be okay to do
<div style="font-weight:bold">
<p>Foo
<p>Bar<span style="font-weight: normal">baz</span>
</div>
to avoid the complexity, given that this sort of markup shouldn't be
too common? I think it's clear that something like "<b>Foo baz
bar</b>" should become"<b>Foo</b>baz<b> bar</b>" and not"<b>Foo
<span style='font-weight: normal'>baz</span> bar</b>", but how far
should we go?
IMO, from the moment you decide to use <b> and not style="bold" (be it
due to a user selectable mode or not), style="bold" should actually be
totally avoided. Thus, the CSS bolding in your example should be
replaced by <b> tags:
<div>
<p><b>Foo</b>
<p><b>Bar</b> baz
</div>
I think that the code generated should be homogeneous, independent from
the original markup (which may be copy-/pasted from various sources).
That means, the bolding in your example would be changed to <b> already
when the contenteditable element is built, or when the fragment is
pasted in.