Post-release news for Aladdin Ghostscript 5.10


This file contains information about bugs found since the 5.10 release, and workarounds or fixes when available.

  • (July 14, 1998) stackunderflow error from PDF files using CalGray color space
  • (June 26, 1998) pdfTeX output gives a typecheck error
  • (June 23, 1998) Patterns don't display
  • (June 4, 1998) "/rangecheck in --get--" when reading PDF file
  • (May 27, 1998) ps2pdf converts 3-sided box to closed box
  • (April 25, 1998) crash when printing Pattern with text operator in PaintProc
  • (March 5, 1998) crash when printing files with dashed lines (setdash)
  • (February 9, 1998) pdf2ps output is incorrect

  • (July 14, 1998)

    Problem: Displaying or printing a PDF file that uses the CalGray color space gives a stackunderflow error.

    Workaround: Apply this patch (does not require recompiling).


    (June 26, 1998)

    Problem: Displaying or printing the output of pdfTeX sometimes gives a "/typecheck in put" error. (This problem is caused by a bug in pdfTeX, not in Ghostscript.)

    Workaround: Apply this patch (does not require recompiling).

    Even though this is not a Ghostscript bug, we will include the workaround in future releases.


    (June 23, 1998)

    Problem: Patterns sometimes don't display.

    Fix: Apply this patch (requires recompiling Ghostscript).


    (June 4, 1998)

    Problem: When processing PDF files produced by recent versions of Adobe Acrobat Distiller, Ghostscript sometimes gets a "/rangecheck in --get--" error with 0 as the top (last) element of the operand stack.

    Fix: Apply this patch (does not require recompiling).


    (May 6, 1998; updated May 27, 1998)

    Problem: When producing PDF output, Ghostscript sometimes adds a fourth (closing) line to stroked rectangles that have one side missing.

    Fix: Apply this patch (requires recompiling Ghostscript).

    Note that this fix may cause ps2pdf to produce larger output for some PostScript files.


    (April 25, 1998)

    Problem: Ghostscript crashes with a memory access error in clist_fill_mask when printing files that contain Patterns whose PaintProc executes a text operator such as 'show'.

    Fix: Apply this patch (requires recompiling Ghostscript).

    Workaround: See Workaround #1 below.


    (March 5, 1998)

    Problem: Ghostscript crashes with a memory access error when printing files that contain any dashed lines or dashed curves.

    Fix: Apply this patch (requires recompiling Ghostscript).

    Workaround #1: Insert the following code into gs_init.ps just before the line that currently begins with /setdash.

       /setdash { 1 index length 0 eq { {1000 0} 0 //setdash pop pop }
          { //setdash } ifelse } odef
    
    Alternatively (not quite as reliable), put the code in a small file (say dashfix.ps), and then mention dashfix.ps on the Ghostscript command line before your own file.

    Workaround #2: If your machine has enough memory (real or virtual), you can tell Ghostscript to allocate a RAM buffer large enough to hold the entire page, with the switch

    	-dBufferSpace=N
    
    where N is at least as large as the result of the following computation:
    	(width_of_page_in_inches * X_resolution * depth + 64)
    	  * height_of_page_in_inches * Y_resolution / 8 + 10000.
    
    The "depth" value is the number of bits of data per pixel in the internal page buffer, and depends on the specific printer or file format. Black-and-white output almost always uses a depth of 1; color output may use a depth of 4, 24, or 32 depending on the driver. For example, a U.S. letter-size page on a 300 DPI color printer using a driver with 32-bit color values would require a minimum N = (8.5 * 300 * 32 + 64) * 11 * 300 / 8 + 10000 = about 33 Mb of buffer space. If you don't know what depth a particular driver uses, try using 4, and if Ghostscript still crashes, try 24 and then 32.

    Workaround #3: You may be able to work around the problem by including the -Z, (dash, Z, comma) switch on the command line, preferably as the first switch; however, this doesn't always work.


    (February 9, 1998)

    Problem: The output from the pdf2ps program is often incorrect.

    Workaround: Instead of

    	pdf2ps xx.pdf yy.ps
    
    use
    	gs -sDEVICE=pswrite -dNOPAUSE -dBATCH -sOutputFile=yy.ps xx.pdf
    
    This produces larger and cruder PostScript files, but they are much more likely to be correct.
    Last edited by L. Peter Deutsch