Re: [Cdrecord-developers] Re: [Cdrecord-video] Re: cdrtools-2.01.01a09 released
items.Hendrik Visage wrote:
On 5/30/06, Joerg Schilling <schilling@fokus.fraunhofer.de> wrote:
Bill Davidsen <davidsen@tmr.com> wrote:
> I wouldn't live in the past if you would get the hint that SCSI is
dead
> for probably 98% of your users, and trying to use connections which
> depent on the fantasy that the devices are scsi is a source of endless
> grief. Only the commands are (sort of) scsi, the transport is almost
> entirely gone for CD/DVD.
Each time you repeat this nonsense, I foster the impression that you are
not only living in the past but that you are also uneducatable :-(
Bill, Although I have issues with the way how Joerg comes over (it
might be a cultural
thing from a person that's typing in a non-mother tongue), I have to
start to agree with
Joerg on this one.
So you think more than a few percent of the CD/DVD users are still using
SCSI as a transport? Or do you agree that the burner interface should
pretend to be SCSI? I have exactly one SCSI burner left, all the rest
have been replace with ATAPI many years ago. Even the protocol isn't
really SCSI, it's a subset with vendor extensions.
Would you please be so kind to inform yourself and discover that the
SCSI
protocol is one of the most sucessfull protocols and that more and
more devices
depend on the SCSI protocol?
Not only this Bill, but that the CD writers still use a SCSI based
protocol (and they
came out initially in SCSI before IDE could handle the *sustained*
throughput).
That's exactly what I meant by "(sort of) scsi" and I used to have a
bunch of SCSI CD burners, with one old Philips 2600 left.
Even worst is that in certain situations I've found that a CD/DVD
writer don't wnat *any*thing
else on the same IDE cable that's trying to talk when that CD/DVD
writer is writing.
Depending on the drive buffer size that may be the case, and it's always
good practive. SATA drives don't even allow physical connection of
multiple devices, and multi-bus controllers are available for people who
need a lot of burners. It should be possible to have multiple burners on
USB, although it's probably desirable to have only one for a controller
there as well.
I have never had a problem with multiple burners on the same PATA cable
as long as I didn't try to use both at once. But I admit I only did that
because of cable issues, not because I think it's a good practive!
If you did, you woul know that ATA is one of many SCSI tranports
(like USB
or 1394 are SCSI transports).
To get back to Linux, the SATA stuff is lately done using the SCSI
transpport layers
with libata inside the SCSI subsytem of Linux :)=)
The fact that Linux does not create an universal umbrella for SCSI
transports
is a hint that the Linux kernel developers are also living in the past.
Note that all modern OS did this already many years ago.
They are getting there, but need a few decent developers that will
focus on
fixing the past legacy of hd<a-z>, not to mention the others, and to
fix the
short comings in the current SCSI subsystem.
If you search the kernel mailing list you will find that I suggested
doing exactly that, and using ide-scsi for everything. and I have a few
old systems which do that because the IDE ZIP drives they need work
better as SCSI. But the kernel didn't move that way, although it may
someday.
No Bill, before I continue to boost Joerg's ego in this regard, please
check in
with the Linux kernel source code, ask Joerg for some
guidelines^H^H^H^H^H^H^H
Rather checkout joerg's webpages in this regard first and then only
ask questions
once you need some clear clarity after other sites confused you, and
help to
fix the Linux subsystems rather than to complain to Joerg.
Since other applications mave manages to cope with Linux as it is in
reality, I can't see the problem as anything but political...
Now if just OpenSolaris had a decent nVidia *and* ATi drivers... but
that's a complaint for
a different forum :(
If it had a larger user base, more applications, a dedication to open
development instead of what appears to be a reluctant reaction to market
forces... or if BSD was an operating system instead of a groups of
slightly incompatible operating systems who can't decide if they
cooperate or compete... but in truth the users have chosen to use Linux,
that's reality vs. wistful thinking.
Historical note: not only is the use of simulated SCSI transport on
non-scsi unfamiliar to most users under 40 who have never seen real
SCSI, but "back when" there were always four numbers, not three. They
were the controller number, often determined by the slot into which the
controller was plugged, then the bus number, then the device number and
the LUN.
--
bill davidsen <davidsen@tmr.com>
CTO TMR Associates, Inc
Doing interesting things with small computers since 1979
Reply to: