Re: LGP2 graphics

[MILLER@vax.cam.nist.gov (Bruce R. Miller)]

    Date: Thu, 1-Mar-90 18:27:34-EST
    Date: Thu,  1 Mar 90 16:19:05 EDT
    From: RUSHING@TITAN.KSC.NASA.GOV

    I've been trying for some time now to get multiple page graphics output
    on an lgp2.  I'm trying to draw lines and have them span across several
    pages, so I can later put the pieces together...
	
	    Now: if I use WITH-HARDCOPY-STREAM wrapped around
    WITH-ROOM-FOR-GRAPHICS and call FORMAT-GRAPH-FROM-ROOT, I get beautiful
    results (huge 10-page graphs, with a little bit of overlap between
    pages); but if I try using GRAPHICS:DRAW-LINE to that stream, the
    machine just cranks away, generating my picture, then dumps a 1 megabyte
    postscript file to the lgp2.  After waiting so long for this, the
    printer spits out a SINGLE clipped page.

	    We don't have the code for format-graph-from-root here, so I
    have no idea how it's done... any clues out there?  Thanx!

				    -Sam Rushing
				     rushing@titan.ksc.nasa.gov

Is it a possibility that rather than actually being clipped by the
software, the printer doesn't have enough memory?

I haven't tried exactly  your case, but  the symptoms remind  me of what
has happened to me a number of  times.  I write notes & articles  in TeX
and include ps graphics generated on the lispm using \special{psfile=...
If there is  too much  TeX &  too many  complicated figures  the printer
close to my office cant handle it; it does just what you've described --
huge files, wait a  long time and  a page or  two comes out  -- no extra
blank pages or  anything. There  is some  kind of  error reporting,  but
because of our  setup it  goes into  a bitbucket.  I can  access another
printer nearby with more internal memory and it succeeds, so I can still
print.

Assuming that this is a reasonable possibility, one might speculate that
FORMAT-GRAPH-.. manages to  output its  stuff relatively  `page-wise' so
that the printer  doesn't have  to hang  onto a  lot of  pending output.
(Think of virtual  memory paging)  I dont  know much  about postscript's
model, but one could imagine that in order to handle a single  draw-line
covering, say, 10 pages, postscript has to have 10 pages worth of bitmap
to write onto?  If this is reasonable, then you could try breaking  your
draw-line into  pieces;  if  is  enough  peices  it perhaps doesn't even
matter  where  you  break  them.    On  the  other  hand,  if  it  isn't
appropriately `merged' into  the graph,  it might  not help  even if  my
speculation is valid.

Bruce
miller@cam.nist.gov