https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=284783

--- Comment #3 from Warner Losh <i...@freebsd.org> ---
git blame says that since Justin's import, we have only 438/4534 lines being
original. But if we look more closely, there's closer to 40 lines that are the
same between the two files that are copyrightable (define names, some structure
names), but even the way we define the structures have completely changed.
IMHO, Justin should have put his standard copyright on this with an
acknowledgement to Julian's original work, and today such a change would be
even more justified. NetBSD hasn't expanded the file as much and retains the
original statement.

Comparing the 386BSD file with the CAM import version shows a lot of changes:

diff -ub scsi_all.h ~/freebsd-cam-import-scsi-all.h | diffstat
 import-scsi-all.h |  836
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------------------------
 1 file changed, 701 insertions(+), 260 deletions(-)

A large part of these changes were to convert from bit fields in the struct
definitions and byte-oriented fields, which I'd rate as a material change (all
the structures changed), plus elimination of nested unions for
scsi_read_capacity and scsi_read_capacity_data. SCSI command numbers and T_
definitions remain basically unchanged (the changes aren't material). Several
#defines were renamed. There were no original comments, apart from the license,
so there's little to no 'creativity' that remains from that original file that
remained in the CAM import, let alone the files evolution since then.

(also note in my first message it's the doctrine of laches, no t, sorry for the
error)

I'll also note one more point that "TFS supplies this software to be publicly
redistributed on the understanding" which can be used to show the intent of TRW
Financial Systems' to make a wide grant. So for the de-minimus bits of this
file that are still around I don't think it's much of an issue. I can trace it
to 386BSD as well, which was also widely distributed with this license. It's no
longer possible to ask TRW Financial Services anything.

I can't find a Mach distrinbution with this code (the 3.0 one I could find is
based on different work, and the 3.5 one is). So I can't document that CMU ever
distributed this code under the CMU license.

Lastly, there never has been a copyright statement in this file that I was able
to find (386BSD even lacks this, as do all the derivitives FreeBSD, NetBSD,
OpenBSD and Dragonfly BSD). So that's actually a larger omission than the
ambiguous grants....

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to