Attachment 5050 Details for Bug 5561 – compute the dashed line slope once and use that slope to compute line faces

[patch] compute the dashed line slope once and use that slope to compute line faces

cairo-line-dash-2nd-try.patch.txt (text/plain), 3.39 KB, created by Jeff Muizelaar on 2006-03-25 16:34:02 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Jeff Muizelaar

Created: 2006-03-25 16:34:02 UTC

Size: 3.39 KB

patch

obsolete

>--- cairo-1.0.2/src/cairo-path-stroke.c	2005-10-03 16:44:43.000000000 -0400
>+++ cairo-1.0.2-0width/src/cairo-path-stroke.c	2006-03-25 00:29:24.000000000 -0500
>@@ -529,6 +529,61 @@
> }
> 
> static cairo_status_t
>+_cairo_stroker_add_sub_edge_slope (cairo_stroker_t *stroker, cairo_point_t *p1, cairo_point_t *p2,
>+				cairo_slope_t *slope,
>+			     cairo_stroke_face_t *start, cairo_stroke_face_t *end)
>+{
>+    cairo_status_t status;
>+    cairo_gstate_t *gstate = stroker->gstate;
>+    cairo_polygon_t polygon;
>+    cairo_slope_t slope_check;
>+
>+    if (p1->x != p2->x || p1->y != p2->y) {
>+	/* XXX: Need to rethink how this case should be handled, (both
>+           here and in _compute_face). The key behavior is that
>+           degenerate paths should draw as much as possible. */
>+    _cairo_slope_init (&slope_check, p1, p2);
>+    if (_cairo_slope_compare (slope, &slope_check))
>+	    printf("slopes don't match\n");
>+//	return CAIRO_STATUS_SUCCESS;
>+	}
>+    _compute_face (p1, slope, gstate, start);
>+
>+    /* XXX: This could be optimized slightly by not calling
>+       _compute_face again but rather  translating the relevant
>+       fields from start. */
>+    _compute_face (p2, slope, gstate, end);
>+
>+    if (p1->x == p2->x && p1->y == p2->y)
>+	    return CAIRO_STATUS_SUCCESS;
>+    /* XXX: I should really check the return value of the
>+       move_to/line_to functions here to catch out of memory
>+       conditions. But since that would be ugly, I'd prefer to add a
>+       status flag to the polygon object that I could check only once
>+       at then end of this sequence, (like we do with cairo_t
>+       already). */
>+    _cairo_polygon_init (&polygon);
>+    _cairo_polygon_move_to (&polygon, &start->cw);
>+    _cairo_polygon_line_to (&polygon, &start->ccw);
>+    _cairo_polygon_line_to (&polygon, &end->ccw);
>+    _cairo_polygon_line_to (&polygon, &end->cw);
>+    _cairo_polygon_close (&polygon);
>+
>+    /* XXX: We can't use tessellate_rectangle as the matrix may have
>+       skewed this into a non-rectangular shape. Perhaps it would be
>+       worth checking the matrix for skew so that the common case
>+       could use the faster tessellate_rectangle rather than
>+       tessellate_polygon? */
>+    status = _cairo_traps_tessellate_polygon (stroker->traps,
>+					      &polygon, CAIRO_FILL_RULE_WINDING);
>+
>+    _cairo_polygon_fini (&polygon);
>+
>+    return status;
>+}
>+
>+
>+static cairo_status_t
> _cairo_stroker_add_sub_edge (cairo_stroker_t *stroker, cairo_point_t *p1, cairo_point_t *p2,
> 			     cairo_stroke_face_t *start, cairo_stroke_face_t *end)
> {
>@@ -649,6 +704,7 @@
>     cairo_status_t status = CAIRO_STATUS_SUCCESS;
>     cairo_stroker_t *stroker = closure;
>     cairo_gstate_t *gstate = stroker->gstate;
>+    cairo_slope_t slope;
>     double mag, remain, tmp;
>     double dx, dy;
>     double dx2, dy2;
>@@ -658,6 +714,8 @@
>     cairo_point_t *p1 = &stroker->current_point;
>     cairo_point_t *p2 = point;
> 
>+    /* compute the slope once for the entire line */
>+    _cairo_slope_init (&slope, p1, p2);
>     if (!stroker->has_current_point)
> 	return _cairo_stroker_move_to (stroker, point);
>     
>@@ -684,8 +742,9 @@
> 	/*
> 	 * XXX simplify this case analysis
> 	 */
>+
> 	if (stroker->dash_on) {
>-	    status = _cairo_stroker_add_sub_edge (stroker, &fd1, &fd2, &sub_start, &sub_end);
>+	    status = _cairo_stroker_add_sub_edge_slope (stroker, &fd1, &fd2, &slope, &sub_start, &sub_end);
> 	    if (status)
> 		return status;
> 	    if (!first) {

Actions: View | Splinter Review

Attachments on bug 5561: 4309 | 4614 | 5050 | 5175

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.