Wakko Warner <[EMAIL PROTECTED]> wrote:
> > An unmodified cdrtools (run as root as documented in the man page)
> > does not have this problem.
>
> 1) I sent this to debian's bug tracking. What was done with it after that
> is not my fault. I found a bug, and explained it.
> 2) The dev=x,y,z part is braindead. As I have noticed, it searches /dev/sgX
> to find x,y,z. Specifying /dev/sr0 or /dev/hdc is much cleaner,
Not correct.
The Linux kernel is broken by design and libscg does its very best to work
around this.
Neither /dev/sr* not /dev/hd* work at all if you use a pre-Linux-2.4 kernel.
For /dev/hd* you even need 2.5 or newer.
I did make proposals for fixing the proken SCSI transport design in Linux
since 1997 but the Linux Kernel developers are nor interested in fixing their
bugs :-(
The only way to access SCSI in Linux-2.2 or older is to use /dev/sg* but
the relation betwenn /dev/sg* and the real hardware is not stable. For
this reason, libscg needs to scan all devices to search for the real device
specified by the natural way of addressing SCSI that is used by libscg:
dev=b,t,l
> especially when one has USB devices and plugs/unplugs. I have always had
> to run -scanbus every single time something on my USB bus changes to use
This is another design bug in Linux. I encourage you to check OpenSolaris
in order to find out that it could be done properly.
> my USB burner again. I have /dev/myburner automatically linked to the
> proper burner device node. No I will not modify for dev=x,y,z as it
> would have been much easier to figure it out from the device itself and
> cdrecord is the ONLY application that uses this that I use.
If you have this done automatically, it should be simple to use the same
machanism to let /etc/default/cdrecord be edited automatically.
> Asides from the dev= stuff, cdrecord, when it works, works great.
>
> I looked through the patch. The only thing I can see is the fact they added
> retrying. The fact that cdrecord (modified or not) searches /dev/sg0
> ... /dev/sgX to find the specified device is what's brain dead. If this
> were not the case, debian's patch would not have made a difference. I wish
What I asume is that the braindead patch tries to add "support" for a broken
removable media manager.
Many recent problem reports for CD/DVD writing are caused by either a broken
kernel or by a user space application that interrupts the writing process,
So it is most likely that these problems are caused by the broken
removable media manager. Your problem is not one of these but this
removable media manager seems to do more than issuing a 'SCSI test unit ready'
command every few seconds so it cannot be called sane.
If an unmodified cdrecord complains about not being able to open devices,
kill this removable media manager and try again.
Jörg
--
EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin
[EMAIL PROTECTED] (uni)
[EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/
URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
