Re: xyouts with p.multi

[Wayne Landsman <landsman(at)mpb.gsfc.nasa.gov>]

> I'm not sure why you would have expected that,
> since I've never heard "normalized" defined by anyone
> with respect to what is *in* a window.

Well,  the reason I keep running into the problem is that if you just
have a single window, the distinction between the plot window and the
device area is unimportant.       Typically, I will have made, say two
plotting programs 'makeplot1.pro' and 'makeplot2.pro' and they will each
work fine, using, say, XYOUTS,0.2,0.8,'Text',/normal internally.    I
will later decide that I want to put both plots on a single page, and so
I naively write

!p.multi = [0,0,2]
makeplot1
makeplot2

and the positioning of the XYOUTS output is completely screwed up.
Now I can compute  the correct positioning using !X.window and /NORMAL
coordinates as David suggested, or using !X.CRANGE and /DATA coordinates
as Liam suggested.      But it does seem to eliminate most of the
advantages of  having access to /Normal coordinates.

--Wayne
Landsman
landsman@mpb.gsfc.nasa.gov