Re: [Plplot-general] RE: Colors (was: no subject)

[<mj...@ga...>]

Alan W. Irwin writes:
 > On 2004-11-13 08:15-0800 Alan W. Irwin wrote:
 > 
 > > On 2004-11-13 00:50-0600 mj...@ga... wrote:
 > >
 > >> 
 > >> As for the widget command syntax, it's:
 > >> 
 > >> <widget> scol0 <i> <color>
 > >> 
 > >> where <color> is any form suitable to XParseColor().
 > >> 
 > >> E.g. from plframe.c --
 > >> 
 > >>    if ( ! XParseColor(plFramePtr->display,
 > >>                       Tk_Colormap(plFramePtr->tkwin), col, &xcol)) {
 > >>        Tcl_AppendResult(interp, "Couldn't parse color ", col,
 > >>                         (char *) NULL);
 > >>        return TCL_ERROR;
 > >>    }
 > >
 > > Is this interpretation also correct for the pure tcl case, i.e., when
 > > you aren't using X and you are using, for example, the postscript device?
 > 
 > I just looked a bit further into this, and plscol0 appears to be defined
 > properly in bindings/tcl/tclgen.c to take 4 character arguments which are
 > converted to integers using atoi.  And if it doesn't have 4 arguments it
 > complains.  So this gets curiouser and curiouser since 4 arguments
 > apparently do not work and the construct (with two arguments)
 > 
 > .f cmd plscol0 0 white
 > 
 > does work. I presume this is one of those cases (which I thought we had
 > sorted out some time ago) where there is a nameclash so the tcl API in
 > tclgen.c is hijacked with something else.
 > 
 > Maurice, once you figure this out, I hope you change things so for all 
 > commands in the tcl interface (and also for the separate plframe interface
 > that is nameclashing with it) a single-colour argument like "white" is
 > accepted (since that is useful) as well as the 3-argument (r g b) form
 > (since that is the expected form if you generalize from the C API).

Known issue, feature-not-bug, etc.

This was a rare case in which I ditched the C API in order to use the common
convention from Tk/X land, where the color is expressed by a name or a hex
string.  If one wanted to, it'd be straightforward enough to support both
syntaxes in the Tcl<->C layer using the # args to choose the right
interpretation.  As for supporting color names from the Tcl bindings, well,
the translation is done by Xlib and you'd have to find some non-X substitute
that is reasonably compatible.

-- 
Maurice LeBrun    mj...@ga...