formatting contexts.
The widths of block formatting contexts depend on the floats prior to
them. To avoid a circular dependency between width assignment and height
assignment, we must guess their widths during the assign-widths pass.
The old code simply used the size of the last float, whether
left-floated or right-floated, but this proved insufficient to handle
layouts like those seen on Reddit. The new heuristic keeps track of both
left and right floats independently and sums the width of all left and
right floats to determine the width of a block formatting context. This
is still insufficient to properly lay out Reddit, but the results are
much more acceptable.
A fully correct approach will require that blocks be laid out again if
the initial guess proved to be incorrect. A `TODO` is in the code to
handle this case.
The float code was old and did not support most of CSS 2.1. So unifying
the two paths both simplifies code and improves functionality.
Improves the Reddit sidebar.
wrapper around them.
Fixes Wikipedia tables leaking out.
Along the way, I refactored tables' width calculation significantly.
This was necessary in order to properly handle floated tables, as some
of the logic had to be ported over from block flows.
construction.
The iteration was incorrect here. Although it accidentally worked
before, it will cause problems when we have incremental style
recalculation.
The `after_block_iteration` reftest will become interesting once we have
incremental style recalc.
Also fixes calculation of background-position percentages:
Rather than multiplying the container size by a percent and aligning the top
left of the image at the resulting width, we also need to subtract a
corresponding percent of the image size, per
http://dev.w3.org/csswg/css2/colors.html#propdef-background-position
"A value pair of '100% 100%' places the lower right corner of the image in the
lower right corner of the padding box. With a value pair of '14% 84%', the
point 14% across and 84% down the image is to be placed at the point 14%
across and 84% down the padding box."
… when parsing as an absolute URL (without a base) fails.
Previously, arguments were parsed as URLs with the current working directory
converted to an URL and used as the base URL. This mostly worked for
relative filenames, except that `?` and `#` had a special meaning
and needed to be percent-encoded.
Fix#3340.
This still needs a lot of work, but it covers the basic
cases and improves wikipedia while passing all existing tests.
Tweak reftest to deal with linux/travis black background.
Add iframe tests to their own subdirectory and add another test case
that used to trigger a fatal error. The new test case uses the
"allow-scripts" sandbox attribute to work around a script task failure
caused by the child frame sharing the same script task as the parent.
