Preliminary BLAST Forum Rules

Everything is public and open. Most of the technical details hammered out in subgroups, which makes the base proposals. The main group discuss these proposals and votes on adopting them. Technical discussions for the most part done on email lists. Subgroup bring their own proposal as handouts. Unified draft put together from these.

For running discussions, use loosely-enforced Robert's Rules of Order. Basically, we won't stand on formalities unless the discussion is becoming unruly. Some of the more often-invoked rules:

• 1. Moving and voting on amendments before the main proposal. Basically, this will keep the discussions focussed.

• 2. Motions coming out of committee (subgroup) are automatically seconded; others need a second from the floor.

• 3. Triply-nested amendments are not allowed. This keeps the confusion down.

Generally, whoever is making the presentation runs the discussion, recognized comments from the floor, etc. Convener comes in after the meat of the discussion to run the votes. Comments during discussion handled by going around the room clockwise.

Will pass around an attendance sheet at each meeting to keep track of who was there. Organizations are asked to commit to having the same attendee at every meeting. It's very important to have this kind of continuity in attendees, else we could have spent too much time in remedial education.

Each organization (school, company, lab) gets one vote - note this is on an organization basis, not a person. Representatives from an organization should agreed on who is voting. An organization is eligible to vote if it had attended 2 of the last 3 meetings, counting the current meeting (i.e. you could attend every other meeting and still vote; you could not vote at your first meeting). Obviously, not enforced at the first two meetings.

Accepting a section of the BLAST spec is a multi-step process.

• 1. Someone writes a draft specification; Details of these are hashed out in the subgroup.

• 2. First Reading: The subgroup leader (or occasionally the draft author) presented the subgroup-approved draft to the main group. When there is controversy, it is usually pointed out. The main group discussed, suggested changes, and holds a series of "straw votes" on the proposal. All attendees are eligible to vote in straw votes, which are not binding on the subgroups.

• 3. More subgroup discussion, both electronic and in person at the next meeting, producing a revised proposal.

• 4. Second Reading: The subgroup leader presented the revised proposal to the main group. Sections that are substantially the same as the original (or an alternative presented at the first reading) are amended by motion and eventually voted on. Eligibility for votes was as explained above. Major additions are treated as first readings at this point. A few sections may be sent back to subgroup for more work; these can come back as second readings at the next meeting.

• 5. Once a section is accepted at second reading, it is "frozen" until the end of the BLAST process. Revisions are only allowed for clarity or when new information surfaced (like discovering that the draft is self-contradictory).

• 6. At the end of the process, we promised to allow reconsideration of any feature.

Subgroups meet independently of the main body, usually the afternoon and evening of the day before. The leader of the subgroup run these meetings using whatever style he or she is comfortable with. Also, there is a mailing list for each subgroup where most of the discussions goes on.

Each subgroup is devoted to one topic from the following list (prelimary):

• BLAS functionality subcommittee

• BLAS lite subcommittee

• sparse BLAS subcommittee

• parallel BLAS subcommittee

• extensions to existing BLAS subcommittee

• data structures subcommittee

• language binding subcommittee

The groups meet in parallel, which may cause a little friction but is logistically unavoidable. When a subject straddled two groups, the subgroup leaders would talk to each other and decide who would handle it - this may lead to minor anti-turf battles (also known as "after you" deadlock). Both groups would act as sanity checks on the results.

When it comes time to write the draft, each subgroup become a chapter. The subgroup leader is the editor (and usually major author) of the chapter, and is responsible for making sure the chapter reflected the decisions made in the subgroup and in committee.

Meetings are 2.5 days, starting Monday or Wednesday afternoon, in Nasville. A typical schedule is

We will try to finish as early as possible on Friday, to allow people to catch flights.

We financed the meetings primarily from a per-meeting fee of $75 per attendee. We have no other source of funding at this time.

For future planning here is a tentative list of dates, roughly 3 months apart, for the series of meetings:

• BLAS session on April 12 at the Copper Mountain meeting

• BLAST Forum Meeting May 20-22 start noon on 5/20 depart noon 5/22.

• BLAST Forum Meeting August 12-14 start noon on 9/12 depart noon 12/14.

• November 18-22 (Supercomputer 96 Meeting)

The Draft: ?? serves as our general editor, collecting the chapters and trying to smooth format details. We will try the following framework:

• 1. The Editor has the "official" version of the draft. He sets deadlines for receiving the chapters, and edits for formatting.

• 2. The Editor sends the Chairman the whole document when all the chapters are done.

• 3. The Chairman sends the document to the "core" mail group (see below), puts it out for anonymous FTP, netlib, and announces it on the net.

• 4. Any further changes are supposed to be made to the Editor's edited copy, not the original version.

Each subgroup writes one chapter, generally written and/or edited by the subgroup leader. Also, the author of each section is identified in a footnote at the beginning of the section, along with the date (and occasionally other version information). The chapter authors should realize they are writing a draft chapter, not a stand-alone document.

Mailing lists: Every subgroup has its own mailing list. Those lists are where most of the technical action happened. On top of that, there is a list for everybody in the world interested in BLAST, and another for the "core" group. The "everybody" list is used for the meeting minutes and a few miscellaneous announcements. The core group list is primarily for the meeting attendees, but a few others are also on it for political and/or practical reasons; it got meeting details, and copies of the various proposals before the meetings. All lists are kept at cs.utk.edu People can add or delete themselves to/from any of the lists by mailing to dongarra@cs.utk.edu.

The following mailing lists have been set up.

• blast-comm@cs.utk.edu Everyone

• blast-core@cs.utk.edu people who attended the last few meetings

• blast-lang@cs.utk.edu Language binding subcommittee

• blast-function@cs.utk.edu Blas functionality

• blast-lite@cs.utk.edu BLAS lite interface

• blast-ob@cs.utk.edu Object based

• blast-parallel@cs.utk.edu parallel processing

• blast-ds@cs.utk.edu data structures

• blast-sparse@cs.utk.edu sparse operations

• blast-ext@cs.utk.edu extensions to the existing BLAS

Minutes: Susan Ostrouchov and Andrew Lumsdaine have agreed to take minutes for the general meeting. Each subcommittee should also have minutes taken.

Announcements of major news (new drafts, etc.) to go everywhere we can think of. This includes the "world" mailing list; newsgroups comp.lang.fortran, comp.lang.misc, and comp.parallel; (indirectly) na-net, scinet, and hpcwire. Meeting minutes go to the world and core mailing lists and the newsgroups. Good idea to mention BLAST when giving talks.

• Jack Dongarra, Convener and Meeting Chair

• ???, Executive Director

• Andrew Lumsdaine, Editor

• Susan Ostrouchov, Minute taker

• Andrew Lumsdaine, Minute taker