25 Jan
2003
25 Jan
'03
2:40 a.m.
Jochen Kunz <jkunz(a)unixag-kl.fh-kl.de> wrote:
Well, I had to use some programm to write the tapes,
dd(1), maketape, or
somthing else. I made good experiences with maketape to write 2.11BSD
tapes, so I stayed with it and it worked very well.=20
[...]
The same for dd(1). There may some implementation differences in dd(1),
say on SunOS or AIX, that may produce unusable tapes. Therefore I used
maketape. I had a glance at the code and it seams that it does
everything proper on every UNIX and Unix-like OS.
Don't forget the chicken-egg problem. If I have no 4.3BSD-Quasijarus
running, I have to use some random foreign OS to produce distribution
tapes. (Or I have to bother somone else to do it for me.)
But dd is a standard general-purpose tool, as opposed to a highly specialized
program for installing 2.11BSD.
I learnd that the bs=3D parameter of dd doesn't
set the block size of the
tape with an ioctl, it is only the buffersize parameter that is used in
the write(2) syscall.=20
On every system I have used the sizes of records written on tapes are
determined precisely by how much you write with one write or writev syscall, no
ioctl needed. But if some weird system does require a special syscall, I can
bet that dd on *that* system will make it. On each system its native dd utility
will always do the right thing, as opposed to some special program ripped out
of a 2.11BSD distribution and used for something it was never intended for (to
write dist tapes for a completely different OS).
MS