Re:[gnuplot-Patches-1353539] Fix clipping for plots with canvas size != 1,1

[Ethan M. <merritt@u.washington.edu>]

> Submitted By: Harald Harders (harders)
> 
> Depending on the terminal, either
> 
> term->xcanvas = term->xmax * xsize;
> term->ycanvas = term->ymax * ysize;
> (when 'set size' changes the canvas)
> 
> or
> 
> term->xcanvas = term->xmax;
> term->ycanvas = term->ymax;
> (when 'set size' does not change the canvas)
> 
> will be appropriate.

In the general case, don't we need to know all four 
corners of the canvas?  So far as I understand the
code in the postscript driver, it uses both 
"set size ..." and "set offset ..." to generate the
true canvas.  For instance:


    switch (ps_params->psformat) {
    case PSTERM_EPS:
        term->xmax = PS_XMAX;
        if (ps_params->oldstyle)
            term->ymax = PS_YMAX_OLDSTYLE;
        else
            term->ymax = PS_YMAX;
        xmin_t = term->xmax * xoffset / (2*PS_SC);
        xmax_t = term->xmax * (xsize + xoffset) / (2*PS_SC);
        ymin_t = term->ymax * yoffset / (2*PS_SC);
        ymax_t = term->ymax * (ysize + yoffset) / (2*PS_SC);
        break;


Don't we really want these same values, or at least their
unscaled equivalents, available to the core code?
E.g.:

	term->canvas.xleft  = xmin_t * PS_SC;
	term->canvas.xright = xmax_t * PS_SC;
	term->canvas.ybot   = ymin_t * PS_SC;
	term->canvas.ytop   = ymax_t * PS_SC;

(give or take a factor of 2).


-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195