** This is the quasi-official and semi-temporary T13 email list server. **
Hale,
We are definitely getting closer to an understanding.
There was an overwhelming desire to get ATAPI into X3T10 as soon as
production started to zoom through the roof in mid-1994.
Pressure came from the ATA-2 working group and queueing/overlap was the
catalyst. There were three proposals on the table at the ATA-3 Study Group
in October 1994 and at the time you said that "Overlapped Commands and
Queueing are the most divisive issues we've got". Incidentally, at the study
group it was noted that one of the features pushed out of ATA-2 was "Include
ATAPI CD ROM support".
The Project Request written for an ATAPI standard included a requirement
that ATAPI conform to SAM (SCSI Architectural Model). After this was
removed, X3T10 approved the project by 38:2 in November 1994.
At November's ATA WG the members were persuaded by Brad Hosler of Intel to
focus on overlap instead of the queueing which had dominated previous
arguments, viz "I need a decision soon enough that I can buy overlap-capable
drives in May next year [1995]".
After X3T10 initiated the ATAPI project, SFF had to be programmed out of the
picture because SFF does not operate rival activities. To reach an end of
life point, SFF-8020 was to continue as the approval body until May 1995.
The 1995 ATAPI working group would meet as an X3T10 activity to integrate
Rev 1.2 and the SFF-8029 Errata specification into a Rev 2.0 that would be
voted on by SFF before handing it off to X3T10 in May.
The schedule was not met. SFF did not see Rev 2.0 before copies of it were
available at X3T10 in March. All hell broke loose because much of WD's
overlap proposal to the ATA WG had been included in it.
The formal handing over of a completed SFF Specification never happened in
May, and since nobody was in a mood to let SFF keep voting control any
longer the transition occurred without the symbol. May's X3T10 was full of
action because the ATA WG pressed for independence. After some heated
discussions it was decided to letter ballot the formation of X3T13.
In the fuss over X3T13 in the next few months, the separate ATAPI project
died. Pete McLean created a brand new alternative to the ATA-3 draft that
merged ATA and ATAPI for the June 1995 ATA WG. The decision was made to
finish ATA-3 without ATAPI, until the next standard which was named ATA+PI
(it later became ATA/ATAPI-4).
That left a vacuum because Rev 1.2/SFF-8029 was woefully out of date, and it
was decided between the players that a new rev of 8020 without the offensive
overlap capabilities would be published as an Information Specification to
serve as an interim document until ATA+PI was started.
> So SFF had nothing to do with the development of the revision 2.x
> documents?
That's right, because none of the activities were held in SFF SSWGs, they
were in X3T10 working groups. There was close reporting back to the SFF
Committee but that was a courtesy thing, not a control factor.
> And I bet I can safely assume T10 did have anything to do
> with them either.
Oh no, it was a formal activity, but control/responsibility shifted in the
period between March and July.
ATA/ATAPI WORKING GROUP MARCH 8-9
Gene Milligan (Seagate) announced that the ATAPI SD3 (Project
Request) had been approved by X3, so the protocol was now a formal
X3T10 activity. The development base will be the SFF-8020 ATAPI CD-
ROM specification plus input from the QIC (1/4" Cartridge) 57
document. And here began a conflict which seemed to excite everyone
in the room.
Tom Hanan (Western Digital) distributed copies of the long-awaited
Rev 2.0 of SFF-8020. Rev 2.0 includes all approved changes to the
base document as well as a first pass at defining the protocol and
register definitions for limited command overlap. The difficulty is
not that Rev 2.0 exists, but that it arrived late: it should have
been submitted to the SFF Committee to vote on so that it could be
forwarded to X3T10 by membership vote.
Although the ATAPI WG had started off in combination with ATA WG activities
it was moved to the SCSI Multi Media Commands WG as you can read here.
MMC WORKING GROUP JULY 11 1995
Devon Worrell (Western Digital) distributed copies of SFF-8020i Rev
2.4.
Yes, this is the selfsame 8020 that has been the subject of so much
overlap-related controversy in recent months. Rev 2.4 was the
compromise agreed to by all parties: remove a repugnant (to disk
drive vendors) advanced overlap but retain all the advanced
commands. The purpose of the review was to have an open editing
pass before producing a stable and final Rev 2.5.
Since few participants in the ATA WG would have attended the MMC, there were
no checks or balances on the ATA/ATAPI protocol section.
A wild card was the scheduling of ATAPI Editor's Meetings to complete the
revisions and these would have been poorly attended.
> OK, I trust you when you say T10 voted to do this. It just sounds a
> little odd to me.
It was a little odd, but then again many of the same players were in SFF and
X3T10. SFF offered a way to achieve an end without taking the time to
complete the standards process. ATA+PI was not going to be started until
1996, and the market needed something in the meantime. Gene reported the
activity to X3T10, so it was hardly hidden from view.
X3T10 MEETING JULY 12 1995
ATA-3: Gene Milligan had quite a bit to say regarding progress in
the ATA-3 working group.
"ATAPI has a new draft being circulated by SFF as an Information
Spec."
"We had a request to merge the MMC (Multi Media Commands) and ATAPI
reflectors and we plan to do that. This will leave us with the ATA,
ATAPI/MMC and SFF reflector sites for support."
"ATA-3 has a new draft ready for distribution and we are looking at
doing an annex to include the SMART (Self Monitoring Analysis and
Reporting Technology) documentation."
> But the problems we are talking about go back to the 1.x documents
> and were never fixed there or in the later 2.x documents.
Since you never raised any of those to the SFF Committee, there was no
vehicle to ensure they were addressed. Seagate was a member and you were
employed there for much of 1994. I have to assume that you chose not to be
public because they may not have been too happy about you 'wasting' your
time on ATAPI.
> No one was responsible for the
> technical content of SFF/INF-8020. BTW the front page of INF-8020
> also says this very same thing.
The wording is intended to help people understand the situation with
Information Specifications.
The history probably should have said something about it being submitted by
the SCSI MMC of X3T10 but that omission was not a big deal at the time
because everybody knew and understood it.
> I do not think I said "SFF was ignoring me".
It was probably a spillover from the rampaging on this reflector about how
T13 ignores technical input and discards ideas from 'outsiders'.
Since SFF runs by a pretty strict set of rules, comments which suggest
otherwise tend to raise hackles.
> What I have always said is that the
> "editors of SFF-8020 ignored me" and I also know they ignored the
> input and comments from many other people.
Editors cannot ignore input if the input is provided in a public forum and
the organization has a capable chair. You were active in ATA at X3T10 during
1995, and a note to Ron Roberts who led the MMC WG or John Lohmeyer as T10
Chair would have focused instant attention on the situation and caused the
editors to respond to your comments.
> It seems that SFF documents are much less hassle to publish
> and they certainly seem to have a longer life span that ANSI
> NCITS documents.
One factor in INF-8020i's long run is that engineers do not have to try and
figure out what to do from several standards.
Another is that Microsoft, Intel, and other major OEMs specifying INF-8020i
in their compliance requirements.
Being specified in RFPs and such is pretty normal for SFF Specifications,
and that is one reason for the 'majority of one' rule. It prevents a spec
ever being Published so long as there is one valid technical negative
comment and although the system is not foolproof, it has been effective in
creating a high level of completeness.
Dal
--
If you have any questions or wish to unsubscribe send a
message to Hale Landis, [EMAIL PROTECTED] To post to
this list server send your message to [EMAIL PROTECTED]
For questions concerning Thistle Grove Industries or TGI's
list services please send email to [EMAIL PROTECTED]
