d.rast.num: Align grid with/without -a preserving the aspect ratio

[HuidaeCho]

This PR addresses #3503 by using the same pattern in d.rast.arrow.

[HuidaeCho]

The second 0 in

grass/display/d.rast.num/main.c

Line 235 in 827f988

D_setup2(0, 0, t, b, l, r);

means don't adjust the destination coordinate system to preserve the aspect ratio.

grass/lib/display/setup.c

Lines 97 to 114 in 827f988

	/*!
	\brief Sets source coordinate system
	Sets the source coordinate system to its arguments, and if
	the <b>fit</b> argument is non-zero, adjusts the destination coordinate
	system to preserve the aspect ratio.
	If <b>clear</b> is true, the frame is cleared (same as running
	<i>d.erase</i>). Otherwise, it is not cleared.
	\param clear non-zero code to clear the frame
	\param fit non-zero code to adjust destination coordinate system
	\param s_top
	\param s_bottom
	\param s_left
	\param s_right
	*/
	void D_setup2(int clear, int fit, double st, double sb, double sl, double sr)

However, setting that argument to 0 only works for the -a flag because the number of cells within the display extent counts actual raster cells, not what's shown with constrained resolution (for no -a flag). This is why d.rast.arrow works fine in d.mon wx0 with/without -a.

[echoix]

Who's knowledgeable to review this, and review this in relation to the other half dozen PRs related to the same problem/fixes? I personally didn't follow after the approach changed in one of the d.* PRs, and I'm lost with the discussions scattered.

[HamishB]

Who's knowledgeable to review this, and review this in relation to the other half dozen PRs related to the same problem/fixes? I personally didn't follow after the approach changed in one of the d.* PRs, and I'm lost with the discussions scattered.

Hi, that's probably me, although I expect @HuidaeCho is a bit more up to speed than I am on the modern display library. But I'm currently refamiliarizing myself with it, admittedly in an effort to understand some low level weirdness I'm seeing in some of my own code as I port the d.barb module from grass6-addons to grass8 (d.barb is a much enhanced version of d.rast.arrow, and partly based on it). I think I'll have to hunt back in the grass-dev archives from August 2008 to see if @glynnc ever fully described the 'new' source → destination rendering philosophy at a high level and if so add that into the Doxygen comments.

n.b. my usual first stop is to revisit the very well written GRASS 5 Programmer's Manual PDF to see how the original library design was intended so at least I have a solid starting point to understand the changes that followed.

regards,
Hamish