gdiplus: bezier test question

[Reece Dunn]

The patch that I have just submitted ("gdiplus: fix the bezier arc
path test (on all platforms).") is a simple fix for the failing test
in graphicspath.c. The failure highlights gdiplus behaviour that is
not directly obvious:

If you have a bezier arc:

static path_test_t arc_path[] = {
    {600.0, 450.0, PathPointTypeStart, 0, 0}, /*0*/
    {600.0, 643.3, PathPointTypeBezier, 0, 0}, /*1*/
    {488.1, 800.0, PathPointTypeBezier, 0, 0}, /*2*/
};

the gdiplus code will make the last entry PathPointTypeBezier |
PathPointTypeCloseSubpath.

For the nature of the existing test, I believe the fix is correct (it
is testing the basic arc path functionality). However, there should be
another test to document the above behaviour.

- Reece

[Dmitry Timoshkov]

"Reece Dunn"  wrote:
> The patch that I have just submitted ("gdiplus: fix the bezier arc
> path test (on all platforms).") is a simple fix for the failing test
> in graphicspath.c. The failure highlights gdiplus behaviour that is
> not directly obvious:
>
> If you have a bezier arc:
>
> static path_test_t arc_path[] = {
>    {600.0, 450.0, PathPointTypeStart, 0, 0}, /*0*/
>    {600.0, 643.3, PathPointTypeBezier, 0, 0}, /*1*/
>    {488.1, 800.0, PathPointTypeBezier, 0, 0}, /*2*/
> };
>
> the gdiplus code will make the last entry PathPointTypeBezier |
> PathPointTypeCloseSubpath.
>
> For the nature of the existing test, I believe the fix is correct (it
> is testing the basic arc path functionality). However, there should be
> another test to document the above behaviour.

This behaviour matches what GDI32 does with adding PT_CLOSEFIGURE.

[Reece Dunn]

On 25/01/2008, Dmitry Timoshkov  wrote:
> "Reece Dunn"  wrote:
> > The patch that I have just submitted ("gdiplus: fix the bezier arc
> > path test (on all platforms).") is a simple fix for the failing test
> > in graphicspath.c. The failure highlights gdiplus behaviour that is
> > not directly obvious:
> >
> > If you have a bezier arc:
> >
> > static path_test_t arc_path[] = {
> >    {600.0, 450.0, PathPointTypeStart, 0, 0}, /*0*/
> >    {600.0, 643.3, PathPointTypeBezier, 0, 0}, /*1*/
> >    {488.1, 800.0, PathPointTypeBezier, 0, 0}, /*2*/
> > };
> >
> > the gdiplus code will make the last entry PathPointTypeBezier |
> > PathPointTypeCloseSubpath.
> >
> > For the nature of the existing test, I believe the fix is correct (it
> > is testing the basic arc path functionality). However, there should be
> > another test to document the above behaviour.
>
> This behaviour matches what GDI32 does with adding PT_CLOSEFIGURE.

Looking into the test a bit more, it is calling GdipAddPathArc to
construct the path, and then ok_path is extracting the path data,
checking that data against the exprected data.

- Reece