[T10] Tape Stream Miroring incorporation into SPC-5 r11 feedback ... "shall direct"
Kevin D Butt
kdbutt at us.ibm.com
Mon Aug 8 09:56:04 PDT 2016
I support not using the word "direct" for the reasons you stated. Your
proposed solution, "the copy manager and the device server shall" is
slightly different from the last sentence which is, "the copy manager and
device server" (i.e., the is only one "the" in use). I may be being a
little finical, but using a "the" in both "the copy manager" and "the
device server" seems to emphasize that they are two separate and distinct
entities, whereas using "the copy manager and device server" does not as
strongly indicate they are different and distinct. In fact, they may be
Maybe I am just being picky, but those are my thoughts. I think your
intent would be met by using the exact phrase in the last sentence, "the
copy manager and device server" instead of "the copy manager and the
Kevin D. Butt
SCSI Architect, Tape Firmware, CAMSS
MS 6TYA, 9000 S. Rita Rd., Tucson, AZ 85744
Fax: 520-799-2723 (T/L:321)
Email address: kdbutt at us.ibm.com
From: Ralph Weber <roweber at ieee.org>
To: Dennis Appleyard <dennis.appleyard at oracle.com>, Kevin D
Butt/Tucson/IBM at IBMUS, curtis.ballard at hpe.com, Darryl Torske
<darryl.torske at quantum.com>, "T10 org (t10 at t10.org)" <t10 at t10.org>
Date: 08/08/2016 02:57
Subject: Re: Tape Stream Miroring incorporation into SPC-5 r11
feedback ... "shall direct"
Regarding the discussion of "the copy manager shall and the device server
shall" detailed below ...
The change currently identified for inclusion in SPC-5 r12 is to delete
the first shall.
This particular change has several advantages.
It is one of the changes proposed during the discussion of the comment
As a one-word change, it is the smallest possible change mentioned during
The change seems to address the oldest outstanding comment, i.e.,
"something is missing".
The resulting text identically matches the last sentence in the cited
paragraph, and that sentence generated no comments.
The suggestion that the original text be reinstated deserves a response,
and that response follows.
There are substantial risks associated with asking any T10 editor to
incorporate the first-ever usage of a particular word, and this was the
initial problem with the phrase "the copy manager shall direct the device
Today, the word "direct" is *never* used as a verb in SPC-5. The only
SPC-5 usage of "direct" is in compound names, e.g., direct-access block
device and direct access memory.
A recourse to non-traditional phrasing typically masks an attempt to
violate some long-standing precept of SCSI standards. With this alert in
mind, finding the technical flaw in the original text becomes a cakewalk.
Great effort has been expended throughout SAM-x and SPC-x to avoid
defining the details of interactions between copy managers and device
servers. For many years, the goal has been to achieve the greatest
possible flexibility of copy manager product designs. Generally speaking,
minimizing explicit requirements on inside-the-device interactions has
been used to achieve this goal.
This decision became the agreed practice after a half-day debate over
whether commands flow through the device server to the copy manager or go
directly to the copy manager. Eventually, it became obvious that the
choices were silence on the subject or using the lunch hour to setup a mud
wrestling pit between the head tables.
As a defense against future letter ballot comments, therefore, the editor
respectfully insists that the only way to restore the original wording is
through a CAP-approved proposal which makes only that change. Should this
event come to pass, the proposal's number (and the vote tally approving
it) will be noted for inclusion in responses to the all-but-certain letter
ballot comments on the sole use of the "direct" verb in SPC-5.
All the best,
On 8/3/2016 9:32 PM, Dennis Appleyard wrote:
Thanks for the review.
My comments are inline below.
On 8/1/2016 1:09 PM, Kevin D Butt wrote:
Ralph, Dennis, and all,
Strictly incorporation issues:
On page 281, fourth paragraph, "then the <<copy manager shall and device
server shall>>establish a
deferred error condition with CHECK CONDITION status" I think there is
something wrong with "copy manager shall and device server shall" The
device server is the one to establish a deferred error. The copy manager
has a requirement to terminate the copy operation which was just stated
prior to the then, so what does the copy manager need to do here? I think,
perhaps, the "copy manager shall and" should be deleted, but perhaps there
was intent to have something here.
I suggest changing to "then the copy manager shall direct the device
server to establish a deferred error condition" This was the wording in
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the T10