The previous code was not handling all cases correctly, (yes,
even something as simple as a quadrilateral can exhibit a
remarkably large number of different cases when tessellation
is attempted).
This fix now introduces slope comparison which handles several
cases that were mis-handled with the previous implementation which
only used independent sorting of the X and Y values of the
coordinates.
This fixes the skew-extreme test case and the bug reported here:
Skew transforms were broken by the cairo update in December
https://bugzilla.mozilla.org/show_bug.cgi?id=373632
This is necessary to ensure that limiting backends using
CAIRO_TEST_TARGET does not increase the number of tests failing,
which is a desirable invariant.
This is more natural since cr->path can be used as if it was a pointer.
This means, for example, if we move on to making it a pointer, most of
the code using it does not need any change. So we get some level of
encapsulation of implementation details, if you prefer the terminology :).
This means, we have to malloc only one buffer, not two. Worst case
is that one always draws curves, which fills the arg (point) buffer
six times faster than op buffer. But that's not a big deal since
each op takes 1 byte, while each point takes 8 bytes. So op space
is cheap to spare, so to speak (about 10% memory waste at worst).
We do this by including an initial op and arg buf in cairo_path_fixed_t,
so for small paths we don't have to alloc those buffers.
The way this is done is a bit unusual. Specifically, using an array of
length one instead of a normal member:
- cairo_path_op_buf_t *op_buf_head;
+ cairo_path_op_buf_t op_buf_head[1];
Has the advantage that read-only use of the buffers does not need any
change as arrays act like pointers syntactically. All manipulation code
however needs to be updates, which the patch supposed does. Still, there
seems to be bugs remaining as cairo-perf quits with a Bad X Request error
with this patch.
Currently the convolution code uses the formula 2*(N-1)-n to reflect the index
n when n is greater than or equal to N.
This is wrong as n=N -> 2*(N-1)-N = N-2 instead of N-1.
Furthermore when the image is small, e.g. at the highest levels of the
pyramid, this causes the code to index before the start of the array and
causes valgrind to issue a warning.
On some architectures, gcc will emit a memcpy for structure copies which will
produce a valgrind warning when the source and destination pointers are the
same. Workaround this issue by explicitly checking the source and destination
for inequality before doing the structure assignment.
