an-1.8

刻录光盘的程序

NEW features of cdrecord-1.8 (shortened):

We are now only a few releases away from the final libscg that includes
SCSI low level command transport.

All:
-	GNU autoconfig support added to the makefilesystem

-	Added PROTOTYPE checking feature

-	Now using autoconf-1.13

-	Moved auto-generation of the include files align.h and avoffset.h into
	the separate directory 'inc'.

	Now all warnings than may be observed with 'gmake' are caused by bugs
	in gmake.

-	Added support for floatingpoint printing on AIX

-	Corrected behaviour for failed commands on non CAM *BSD code

-	New rules for IRIX 64

-	Reduce maxdma to 64k-1 on SunOS 4.x because esp hardware on VME boards
	does not support more than 16 bits for the DMA counter.

-	Allow 16 SCSI busses and 16 targets per SCSI bus if possible with
	an a specific OS.

-	Linux Parallel Port driver now searches backwards for a bus slot.
	This was needed because if SCSI_IOCTL_GET_BUS_NUMBER works,
	there may be holes in the bus mapping array.

-	Linux scsi-linux-sg.c fix for "Cannot send SCSI cmd via ioctl"
	message with cdrecord -scanbus. Fix for sg_mapbus() function.
	This was needed because if SCSI_IOCTL_GET_BUS_NUMBER works,
	there may be holes in the bus mapping array.

-	Better handling for the SCSI status byte with the enhanced Linux
	sg driver interface.
	Look for:
		ftp://ftp.fokus.gmd.de/pub/unix/cdrecord/alpha/SG-2.2.tar.gz (Linux-2.2)
		ftp://ftp.fokus.gmd.de/pub/unix/cdrecord/alpha/SG.patch.tar.gz (Linux-2.0 & 2.1)

	It seems that the Linux kernel people don't like to support
	cdrecord and changed the interface in a way that doesn't allow
	above patches to work in newer kernels.

-	Use Bit field order definitions from previous autoconf run
	rather than using static definitions.

-	Added a new type __SBOOL to be able to deal with BOOL definition
	from Windows.h

-	Added support for compiling cdrecord on SCO Openserver.
	Using usleep() from libc would kill the process with SIGALRM.
	We now simulate usleep() with select().

-	This is the first source release of cdrtools that support 
	WNT / W98 / W95.

	Cdrtools has no known limitations on Win32.

	The FIFO code from cdrecord and cdda2wav is fully supported,
	even 'mkisofs | cdrecord' works. If you want to make multi-session
	CD's on WIn32, you need to use the SCSI code os the mkisofs
	from the cdrtools package. Use mkisofs -M 0,6,0 to access
	a drive on target 6.

-	Makefiles support for 9000-780-hp-ux, 9000-782-hp-ux, sparc64-linux

-	Fixed probems with Linux Parallel port SCSI driver which were caused
	by not clearing the CDB past 6 byte commands.
	Thanks to:
		 Eduardo Trapani <[e-mail]>

-	First "complete" low level implementation of libscg
	Everything that is needed to call scsicmd() is now inside this
	library.

-	floatingpoint support for new glibc working again
	(workaround for not working gcvt() from newer glibc implementations)

	... Yes, glibc2 has a gcvt() but it calls sprintf() to do its work.
	Sometimes sprintf() from glibc gets called then it worked, sometimes
	when the local sprintf() gets called it does not work. This incorrect
	behavior is caused by a bug in the GNU library design and by a bug in
	the the dynamic linker.

-	Forgot to mention that the fixed probems with Linux Parallel port SCSI driver
	which were caused by not clearing the CDB past 6 byte commands.
	(fixed in 1.8a23)
	I haver to thank:
		 Eduardo Trapani <[e-mail]>

-	New SCSI->local pointer to hold all local data.
	This should make libscg now reentrant.

-	Moved routines from port.c to libschily

-	New function scsi_initiator_id() added to libscg

-	libscg now gives much better error messages for open errors
	This is again a step towards the separate libscg.

-	Workaround for a bug in fcvt() on Cygwin32. libschily now works
	again on cygwin32

-	Now 2 years after I created it, scsi-next.c is publically available.
	This makes it possible for everyone to compile cdrecord on 

		- NextStep
		- Apple Rhapsody
		- Apple MacOS X

-	libscg now got official version number 0.1 and allows version checking

-	several macros to convert integers in network byte order to machine
	byte order modified to make them correct on 64 bit machines too.

-	New supported architecture 9000-777-hp-ux

-	Added support for the new smake source distribution.

-	Added new AUTOMAKE support for the makefile system when using smake.
	The makefile system will now support compilation on unknown platforms
	if compilation is done with smake.

-	install-sh script will now deal with -o/-g flag as expected
	We now install with predefined user and group

-	NetBSD does not define unix !!! Added to makefile system

-	configure now works on OS/2 too

-	scsi-os2.c changed for better OS/2 support
	(see  Changelog from Chris Wohlgemuth)

-	An attepmt to allow compilation on HP-UX using the non ANSI
	C-compiler

-	Linux pg.h include file now available as local copy.
	This will make cdrecord compilations work even when the machine did not
	had PP support while cdrecord has been compiled.

-	New 'slot' feature for the makefile system allows easy merging
	of partial source trees.

-	First attempt to support QNX in the makefile system.

-	Legacy support for --longopt
	This is active for all command that use getargs() from libschily.

-	Finally the complete suite may be compiled with K&R compilers.
	This allows cdrecord to be compiled using the free compiler
	that comes with HP-UX.

	To compile with this compiler call:

		gmake CCOM=cc 'COPTS= ' 

-	Support for Amiga NetBSD

-	Support for HP-9000-899

-	autoconf code now checks for _filbuf() to allow compilation
	on HP-UX-11.x

-	align.h now uses (UIntptr_t) casts to avoid GCC warnings on 64bit
	machines. Note that this is just a cosmetic operation as the GCC
	warning was wrong.

-	autoconf check for va_list == ARRAY corrected to allow compilation
	on QNX

-	compiles again on OpenBSD
	The automatic bitorder recognition had some typos....

-	Support for alpha-NetBSD

-	First support for compilation on QNX
	Note that there is currently no SCSI transport code.

-	Changelog ready for 1.8

-	README ready for 1.8

-	Enhanced README.ATAPI

-	Enhanced README.aix by hints on how to create bootable CD's on AIX

-	Support for HP-9000-820


Cdrecord:

-	-toc behaviour corrected for drives that are jumpered to 512 byte
	sector size.

-	Sector size is reset to the old value after writing a CD. This
	help to use a CD-R/CD-RW drive on Solaris that is jumpered to 512 bytes.

-	Starting with official support for SCO Openserver 5

		NOTE: You will need to have a CD in the closed tray of the 
		drive otherwise cdrecord is not able to open the device.

-	Fixed the low level error handling code for SCO Openserver.

-	Fixed the autoconf detection stuff for major()/minor()

-	Fixed the autoconf detection stuff for not working mlockall() on SCO/HP-UX

-	New S_IS*() macros to allow compilation on OS/2 with missing S_IFBLK

-	Added a static configuration for VMS

-	Added a file cdrecord/build_all.com to compile cdrecord on VMS

-	Fixed isnan() code to fit SCO Openserver and VMS

-	New default file handling (/etc/default/cdrecord)
	See new man page and the file cdrecord.dfl in the
	cdrecord directory.

	This allows e.g. cdrecord dev=teac

	To specify the TEAC drive.

-	First attempt to support BeOS

-	New option -force allows to blank defective CD-RW's
	that otherwise would cause cdrecord to complain about
	a media with non recoverable TOC errors.

-	Fixed a bug with ISRC's in DAO mode.

-	Fixed a bug with non page aligned buffers if the FIFO was
	present. This caused problems with FreeBSD/current/CAM
	due to the fact that the max. number of DMA pages was 
	exceeded.

-	Add code to read back the used writing speed from MMC drives.
	This allows to call cdrecord speed=4 on a CD-RW and to read
	Starting to write CD/DVD at speed 2 ....

-	Added support for -toc on the Philips CDD-522

-	automatic switch to -audio if a .au or .wav file was found

-	experimental flag -force to help blanking bad CD-RW disks.
	Use cdrecord -force blank=all of cdrecord -force blank=fast
	after waiting 2 minutes after inserting the bad disk into the drive.

-	DAO now works for Yamaha 4416 in non dummy mode for CD-R media too.

-	new experimental option -useinfo currently disabled in published version.

-	Cdrecord nor reads CD-Text if called cdrecord -toc -vv 
	and writes the data into a file cdtext.dat in the current directory.
	This is a preparation for CD-Text copy features.

-	Partial ATIP reading and full CD-R manufaturer recocgnition
	support for Sony CDU 920/924/926/948

-	Better DAO integration: cdrecord now prints an error message for
	drives that cannot do DAO or if cdrecord does not yet implement DAO
	fot this drive.

-	Cdrecord now looks for 16 SCSI busses and 16 targets per bus.
	Note that not all operating systems will support wide SCSI busses.
	For Solaris, you will wait for a new scg driver which will
	be available soon.

-	Fixes in several source files to make the compiler not 
	complain about signed / unsigned comparison.

-	Fix for Sony CDU-920 ATIP printing. This drive does return
	a ATIP lead-in start of 0!

-	Try to handle the SCSI get mode data Firmware bug which can
	be observed with JVC 2040 ?? (CD-RW) and the TEAC 532
	where mode page 0x2a returns garbage data.

-	Restore sector size if the 10 second wait time
	is aborted by ^C.

-	Added driver object routine to read buffer capacity.

-	New option -atip will print ATIP as a stand alone task.

-	pr_manufacturer()/disk_rcap() now uses rw and audio parameter
	to distinguish between metal and organic dye

-	Old Philips driver now prints Restricted/Unrestricted ATIP info too.

-	Trying to clear MMC drive status after a possible -dummy run.
	This helps to avoid reloading the media after a -dummy run.
	But it does not work with all drives.

-	Philips CDD-2xxx driver now distinguishes between audio and data disks

-	Better Priority handling for OS/2

-	Added wildcart expansion for EMX on OS/2 and non POSIX shells

-	Now setting STDIN to O_BINARY if reading from stdin.
	This allows 
			mkisofs | cdrecord

	to work correctly on WNT and OS/2

-	For DAO, fixed pregap handling for mixed mode CD's where the 
	first track is a data track.

-	Help Adrian Schroeter <[e-mail]> http://members.tripod.com/~fehlfarben 
	by adding inquiry code for thr Ricoh RO-1060C
 
-	better man page entry for -pad option

-	Pioneer DW-S114X now uses the SWAPAUDIO flag in configuration

-	Handle non compliant FAKE inquiry data that MS NT-4.0 creates
	for IDE disks in ATAPI emulation.

-	Handle the fact that the Ricoh 7040 returns NOT READY for
	Mode sense page 2A while loading as disk.

-	Trying to make cdrecord silent on fixating with new MITSUMI ATAPI drives.

-	-msinfo option handling fixed.
	Now returns correct values even if drive is in DAO mode.

-	FIFO timeout enhanced to be more than the SAO write timeout.

-	Workaround for a firmware bug in the Ricoh MP 7040 and MP 7060
	Both drives return wrong data for the first writable address in DAO mode 

-	Attempt to support real time priorities on Win32 - not yet tested.

-	Added shared memory support (FIFO) for OS/2

-	Now prints "HOST ADAPTOR" in the -scanbus output if the OS supports
	to retrieve the SCSI initiator ID

-	New cdrecord -scanbus format includes dev= syntax

	        1,4,0   4) 'PLEXTOR ' 'CD-R   PX-W8220T' '1.00' Removable CD-ROM

-	Added a missing include file for the new realtime pri code
	to make cdrecord compile again on cygwin32.

-	better output text for disk type

-	Now after more then half a year of testing,
	auinfo.c is included with this distribution.
	This allows to create bitwise correct audio copies:

	cdda2wav -B -v255
	cdrecord -v -dao -useinfo *.wav

	When using these commands, the cddb ID is copied, 
	also the start time of the first track, all available
	ISRC ID's, the MCN, the size of all pregaps between
	all tracks and all indices if present.

-	prints warning if an unofficial version of libscg is used

-	removed a bug that caused an endless loop if a TEAC-CDR 50 /55
	was present without a media.
	This bug has been introduced when tryint to find a workaround
	for a bug in the Ricoh 7040 firmware.

-	The Optima Dis Covery 650 CD-R is now recognised as Sony-CDU-924

-	Fixed a bug in the size computation that caused the media to loos 150
	sectors too large if cdrecord was run in DAO mode.

-	New updated macros in intcvt.h make cdrecord more correct on 64 bit 
	machines.

-	Next_writable Address disabled for CD-ROM drives.
	This will abort is someone tries to do a cdrecord -msinfo on a CD-ROM
	drive. Before, this resulted in bad values now it results in an 
	error message.

-	Sector size should now be restored in any case. Added 5 new cases.

-	added forgotten code to make the autodetection for the Kodak-PCD-600
	work.

-	kill fifo background process if cdrecord dies from signal or
	error message.

-	Sector size is now set to 2048 bytes earlier to make 
	cdrecord -msinfo work right.

-	Exit code reworked to make sure that sector size is always set back
	and the tray is unlocked on all error situations.

-	Version information now includes CPU Vendor and OS info

-	Being unable to lock memory is no more considered a fatal error.

-	Being unable set real time priority is no more considered a fatal error.

-	New help message for audio files that are not a multiple of
	1/75 in length

-	include <sys/procset.h>, Needed for SCO Openserver and SVSv Real time priority

-	CDR_DEVICE now correctly superseedes the default in /etc/default/cdrecord

-	Make cdrecord -load only print one error message when the
	tray is empty.


Cdda2wav:

-	cdda2wav now uses mmap() in favour of shmat() on systems that
	support to get shared memory with mmap()

-	cdda2wav now uses pipes instead of SV semaphores.

-	Recognition of FreeBSD -> reducing shared mem to < 128K 

-	Linux: Support of explicit SCSI-Device: -D/dev/sg1:3,0 
 
-	FreeBSD now has CD Extra Support (display of titles) 
 
-	Assembler routines removed (resample.c) 
 
-	New Frontends and OtherProgs with hints
 
-	real time scheduling (SysV rtprioctl) tested with SCO Open server 5.0

-	the scsi commands use page aligned transfer buffers
	(Buffer handling changed to avoid problems on FreeBSD-3.x/CAM)

-	changed the synchronization method of overlapping sectors
	       (low jitter drives should work with -P1)

-	fixed bug with devices like /dev/sga

-	reintroduced signal handlers to do clean up work on signals

-	first support for OSs, which cannot do fork() (depends on HAVE_FORK)