display; HTML does not validate
There are actually 3 different bugs reported in 300809. This checkin fixes
the problem where the spacing and the image are too close: <P></P><BR><IMG>
edtbuf.cpp:
- Modified CEditBuffer::ParseOpenTag() so we don't discard a <BR> that
comes after a </P>.
- Commented out code that adds a blank container before a table that
follows a </P>.
edtele.cpp:
- Modified CEditContainerElement::AdjustContainers() to add a blank
container after a paragraph that ends with a </P> tag.
We redo the way tables get freed.
Earlier when the LO_TABLE element got freed, it would free stuff that hangs
of the subsequent LO_CELL elements. Now, only table related stuff gets freed
and the LO_CELL peer data structures get freed when the LO_CELL is freed.
Lots of code that was if-defed or commented out since Mariner hit the tip
has been removed to improve code readability.
and lo_BoundaryJumpingNext() for the editor context.
If you don't do it the editor, backspacing between lines gets hosed.
If you do it for the browser, selection goes into an infinite loop.
Fix was reviewed by rjc and dp when it went into the 4.06 branch.
Fixed the case where an unclosed single quote in a
tag causes the STYLE attribute to not be found. This was causing
an infinite loop where layout would process the style attribute repeatedly.
See bug report for details. Affected code is inside lo_ProcessStyleAttribute().
Now text layout looks at both the state structure and the current text block
for propagating anchor information when a new text layout element is created.
loop by letting us break out on a condition that wasn't being checked
earlier in lo_LayoutTextBlock() in laytext.c. Thanks a lot to Jeff
Caldwell (look@shell.lazerlink.net) for the patch.
relayout of normal text as well.) Make sure we start at the element
immediately following the LO_TextBlock (if possible, otherwise start
at the startTextElement.) Also, loop up to and including the
endElement. Lastly, insert new soft line breaks after recycling the
old ones.
getting heights of 10000 pixels. Thanks to Tim Rowley (tor@cs.brown.edu)
for providing a patch that was very close to the real fix. You saved me
a lot of work!