Originally Posted by user_none
This is only correct if the file does not use named entities.
That's sort of my point. Bob's test ePub didn't use
any named entities (with the exception of the & amp; which is supported), only the unicode non-breaking space character
. So since the ePub didn't contain any unsupported entities, the DOCTYPE should have been unnecessary. The file really only needed the DOCTYPE after
Sigil changed the unicode characters to named entities upon opening (to avoid Qt changing it to a normal space). That's the part that feels a bit wonky to me. Not that I know how else it could be handled to preserve the intent of the non-breaking space character when cleaning is turned off, mind you.