<html><head><style type="text/css"></style></head><body><div style="font-family:times new roman, new york, times, serif;font-size:12pt"><div style="font-family: times new roman,new york,times,serif; font-size: 12pt;">So if  one of us  develop a new cairo function, let's say draw_chord() that achieves the functionnalities below (like drawing an ellipse without touching the transformation matrix) with an additional radius_y parameter like : <pre class="programlisting">void cairo_chord(<a href="http://www.cairographics.org/manual/cairo-cairo-t.html#cairo-t">cairo_t</a> *cr, double xc, double yc, double radius_x, double radius_y, double angle1, double angle2);</pre>Would there be chances that it is accepted as a contrib, and would there be special recommandations to follow ? (i.e: adding a draw_chord_negative() api as well to stay orthogonal with previous arc api) Fabien <div style="font-family: times new roman,new york,times,serif; font-size: 12pt;">----- Message d'origine ---- De : Behdad Esfahbod <behdad@behdad.org> À : Fabien Costantini <fab672000@yahoo.fr> Envoyé le : Lundi, 10 Juillet 2006, 5h01mn 11s Objet : Re: [cairo] cairo arc drawing limitation <div>On Mon, 2006-07-10 at 06:21 -0400, Fabien Costantini wrote: > Hi I just dived in documentation, and noticed that the only way to get an ellipse > path drawn seems to use the transformation matrix. > > This approach, though at first sight elegant, seems quite limitative to me, in effect : > > 1. To achieve a scale, you have to know what is the current scaling policies, in particular, when dimensions are not in the [0,1] space but in standard pixel width,height coordinates;  it implies that you keep track of the width and height of your current drawing surface to achieve aa correct scale. Also a matrix translate accumulation has to be done (and possibly the inverse translate after the operation) > > 2. most of the time you have to save and restore the context, that could be be avoided > > Wouldn't it be faster to run and simpler to code > just to add one parameter to specify a height possibly different from the width in cairo_arc() > andd cairo_negative_arc() ? > > It would then be more orthogonal with other path construction API like  cairo_rectangle() > (there is fortunately no need to use a cairo_square() function and a transformation matrix to draw a rectangle) > > Any thoughts ? I tend to agree.  It's just too hard to draw a rectangle with rounded corners that are parts of an ellipse. One more thing that comes to mind is the pen width discussion.  Without locked pen width, it's a bit limiting to use cairo_scale to draw ellipses. behdad > Fabien </div></div> </div></div></body></html>