In ESP Ghostscript many bugs got fixed, but probably these bugs and fixes never got reported to upstream Ghostscript. So if distributions switch to the new merged GPL Ghostscript several bugs will appear again. This bug report is a list of such possible regressions and has all patches (from ESP GS) attached. Here are the bugs: http://www.cups.org/str.php?L261 ps2epsi fails when invoking sed http://www.cups.org/str.php?L1079 Running gs causes segfault on Linux/ppc http://www.cups.org/str.php?L1080 "pxlmono" (PCL 6/XL) driver prints CUPS test page with black background http://www.cups.org/str.php?L1100 Output from pswrite cannot be printed to Epson PS printer http://www.cups.org/str.php?L1125 using full path in scripts like pdf2ps etc. ? http://www.cups.org/str.php?L1235 Function `malloc' implicitly converted to pointer http://www.cups.org/str.php?L1399 Adds error-check to pdfopt http://www.cups.org/str.php?L1419 "gv" does not work with the "gsx" of ESP GhostScript with shared libgs http://www.cups.org/str.php?L1433 ps2pdf does not work with the "gsx" of ESP GhostScript with shared libgs http://www.cups.org/str.php?L1455 TTF glyphs outside encoding vector are expanded in outlines http://www.cups.org/str.php?L1539 "ltime" has no linkage http://www.cups.org/str.php?L1549 Error: ./src/zdevcal.c, line 47: tm_year" is not a member of "((struct tm ...)0)" http://www.cups.org/str.php?L1570 Emacs ps-print files no longer work in gs http://www.cups.org/str.php?L1577 bjc600's gs_closedevice breaks get_page_device and causes NumCopies to barf http://www.cups.org/str.php?L1630 ps2epsi: insecure temp files http://www.cups.org/str.php?L1643 Fix sed usage in ps2epsi http://www.cups.org/str.php?L1779 pdf2ps failed on a PDF with the following error : Warning: File has insufficient data for an image http://www.cups.org/str.php?L1635 Ubuntu patch to fix coredumps http://www.cups.org/str.php?L1641 Prevent pdf2ps generating "null setpagesize" http://www.cups.org/str.php?L1794 Wrong page size with laserjet.ppd http://www.cups.org/str.php?L1987 Duplex support in pstoraster is not working properly http://www.cups.org/str.php?L2027 dvipdf does not pass -R flag to dvips For these bugs there is no STR at cups.org: Revision 146: Fixes of ampersand use in src/gdevps.c and src/scfd.c Revision 154: Fix handling of filenames containing spaces Revision 155: Fix debug printf's so they have newlines Revision 156: The pswrite device also needs the CRD stuff Revision 167: Don't use AC_FUNC_MALLOC test, that causes the rpl_malloc Revision 172: Embed C strings and not an array of character values, as the HP-UX C compiler can't handle such a large array without running out of memory Revision 197: Add CUPS fonts directory to default font path All these bugs are fixed in ESP Ghostscript 8.15.4 (patches from ESP GS attached). I did not test whether they are also fixed in GPL Ghostsccript. So please test and overtake fixes which are not yet done in GPL Ghostscript.
Created attachment 3105 [details] Running gs causes segfault on Linux/ppc
Created attachment 3106 [details] "pxlmono" (PCL 6/XL) driver prints CUPS test page with black background
Created attachment 3107 [details] ps2epsi fails when invoking sed
Created attachment 3108 [details] Output from pswrite cannot be printed to Epson PS printer
Created attachment 3109 [details] using full path in scripts like pdf2ps etc. ?
Created attachment 3110 [details] Function `malloc' implicitly converted to pointer
Created attachment 3111 [details] TTF glyphs outside encoding vector are expanded in outlines
Created attachment 3112 [details] "gv" does not work with the "gsx" of ESP GhostScript with shared libgs, ps2pdf does not work with the "gsx" of ESP GhostScript with shared libgs, "ltime" has no linkage, Error: ./src/zdevcal.c, ...
Created attachment 3113 [details] Adds error-check to pdfopt
Created attachment 3114 [details] bjc600's gs_closedevice breaks get_page_device and causes NumCopies to barf
Created attachment 3115 [details] Fixes of ampersand use in src/gdevps.c and src/scfd.c
Created attachment 3116 [details] ps2epsi: insecure temp files
Created attachment 3117 [details] Fix sed usage in ps2epsi
Created attachment 3118 [details] Fix handling of filenames containing spaces
Created attachment 3119 [details] Fix debug printf's so they have newlines
Created attachment 3120 [details] The pswrite device also needs the CRD stuff
Created attachment 3121 [details] Don't use AC_FUNC_MALLOC test, that causes the rpl_malloc
Created attachment 3122 [details] Embed C strings and not an array of character values, as the HP-UX C compiler can't handle such a large array without running out of memory
Created attachment 3123 [details] pdf2ps failed on a PDF with the following error : Warning: File has insufficient data for an image
Created attachment 3124 [details] Ubuntu patch to fix coredumps
Created attachment 3125 [details] Prevent pdf2ps generating "null setpagesize"
Created attachment 3126 [details] Duplex support in pstoraster is not working properly
Created attachment 3127 [details] Wrong page size with laserjet.ppd
Created attachment 3128 [details] dvipdf does not pass -R flag to dvips
Created attachment 3129 [details] Add CUPS fonts directory to default font path
Some test results: http://www.cups.org/str.php?L261 ps2epsi fails when invoking sed -> FIXED http://www.cups.org/str.php?L1079 Running gs causes segfault on Linux/ppc -> FIXED http://www.cups.org/str.php?L1080 "pxlmono" (PCL 6/XL) driver prints CUPS test page with black background -> FIXED http://www.cups.org/str.php?L1100 Output from pswrite cannot be printed to Epson PS printer -> STILL PRESENT Do cat /usr/share/cups/data/testprint.ps | gs -q -dBATCH -dPARANOIDSAFER -dQUIET -dNOPAUSE -sDEVICE=pswrite -sOutputFile=x.ps - and x.ps will contain "end readonly def". http://www.cups.org/str.php?L1125 using full path in scripts like pdf2ps etc. ? -> STILL PRESENT
Let's mark the patches that has been applied to the HEAD or not needed for some other reason as "obsolete".
http://www.cups.org/str.php?L1235 Function `malloc' implicitly converted to pointer -> STILL PRESENT There is still no "#include <stdlib.h>" in src/fapi_ft.c. http://www.cups.org/str.php?L1433 ps2pdf does not work with the "gsx" of ESP GhostScript with shared libgs -> STILL PRESENT http://www.cups.org/str.php?L1419 "gv" does not work with the "gsx" of ESP GhostScript with shared libgs -> FIXED gv works with gsx. http://www.cups.org/str.php?L1433 ps2pdf does not work with the "gsx" of ESP GhostScript with shared libgs -> FIXED test.pdf is the same for both gsc and gsx. http://www.cups.org/str.php?L1455 TTF glyphs outside encoding vector are expanded in outlines -> FIXED Bug #688017
http://www.cups.org/str.php?L1539 "ltime" has no linkage -> FIXED http://www.cups.org/str.php?L1549 Error: ./src/zdevcal.c, line 47: tm_year" is not a member of "((struct tm ...)0)" -> FIXED http://www.cups.org/str.php?L1570 Emacs ps-print files no longer work in gs -> STILL PRESENT (ManualFeed patch from attachment #3112 [details] not applied) http://www.cups.org/str.php?L1577 bjc600's gs_closedevice breaks get_page_device and causes NumCopies to barf -> STILL PRESENT http://www.cups.org/str.php?L1630 ps2epsi: insecure temp files -> STILL PRESENT If mktemp is not available everywhere check if it is available and use it if so.
Comment on attachment 3125 [details] Prevent pdf2ps generating "null setpagesize" The attachment #3125 [details] should not be committed. There's nothing wrong with "null setpagesize". setpagesize is a procedure than can take care of its arguments. The patch effectively disables all custom page sizes.
Comment on attachment 3115 [details] Fixes of ampersand use in src/gdevps.c and src/scfd.c Committed as a rev. 8096
http://www.cups.org/str.php?L1643 Fix sed usage in ps2epsi -> STILL PRESENT http://www.cups.org/str.php?L1779 pdf2ps failed on a PDF with the following error : Warning: File has insufficient data for an image -> STILL PRESENT http://www.cups.org/str.php?L1635 Ubuntu patch to fix coredumps -> STILL PRESENT http://www.cups.org/str.php?L1641 Prevent pdf2ps generating "null setpagesize" -> NOT A BUG http://www.cups.org/str.php?L1794 Wrong page size with laserjet.ppd -> STILL PRESENT
Please open individual bug reports for problems that are remaining.
Comment on attachment 3108 [details] Output from pswrite cannot be printed to Epson PS printer The patch is committed as a rev. 8097.
For all remaining issues separate bug reports are posted now.