Weekly Planning Meeting, 6/12

[Marc Stiegler <marcs>]

Highlights 6/12:

Well, a lot of things happened this week, mostly very good, though 
not all.

The humongous congratulations this week go to markm; for perhaps 
the first time in human history, Mark Miller is ahead of schedule 
:-) Markm completed endorsement tracking this week. Moreover, 
in his spare time he resolved all remaining problems with bootstrapping 
the permissions system--a set of problems for which we had allotted 
a month because of the number of uncertainties. Some of the credit 
for this breakthrough goes to ravi, who had an insight that greatly 
simplified the problems involved. In any case, getting a month's 
jump on permissions is particularly exciting because this is 
one of the two critical paths to Beta. 

Big congratulations go to dean, roger and roland, ravi, and bobp. 
Dean finished implementing and testing 7 modules in the ent-to-disk 
system, which is the other critical path (in addition to permissions) 
for the server. 

Roland has put a lunch bet on formic (with the caveat that he 
gets to fix the bug rick has discovered before invoking the bet 
:-) Ravi has Documents and Links running over the LAN under C++. 
Roger has inlines working (with some of the credit going to roland 
and markm, who played important roles).

Bobp has gone into draft 2 of the Simplified Xanadu Interface, 
the interface to Xanadu we're counting on to allow normal programmers 
to use Xanadu ("normal programmers" are distinguished from "real 
programmers" here; "real programmers" would of course relish 
the 1000+ subroutine calls that the novice must currently memorize 
before starting work :-) Bobp's draft has 37 subroutines; when 
fully fleshed out, there will be less than 100 subroutines, and 
there is a natural, easy-to-understand layering even there (even 
a software engineering manager should be able to understand them--at 
least, I can :-)

The big condolences this week go to hugh, who encountered a severe 
bug in the Mac environment: he ran the linker for 14 hours on 
the Xanadu object libraries before shutting it down. This had 
happened several times already; the difference this time was 
that hugh was quite sure that we were not in any way at fault 
ourselves. Thus, this is a pretty intractable problem (short 
of more exotic solutions, such as the AutoCAD tugboat).

Fortunately, in late-breaking news, a special thanks for service 
and support goes to someone I never would have guessed would 
earn it: Apple Computer. 

Is it possible? Good service support from Apple Computer? Naaw! 
But it's true: within 24 hours of submitting a bug report on 
this problem, the woman in charge of the MPW Linker was here 
on site to find out what was wrong. As you might guess, Apple 
has never seen a set of object files as big as ours--"you beat 
Claris all to pieces" was her comment (Claris has a program that 
takes 2 hours to link). We don't know what's wrong yet, but at 
least Apple is taking the problem very seriously. Meanwhile, 
hugh has plenty of other things he can work on :-)

Meetings:  Dean, markm, stamps, iobjects, etc. Hill, michael, 
checkpointers. Markm, dean,  discuss booting. Markm, dean, ravi, 
restarts. Marcs, rick, Montage requirements for Mac expertise. 
Hugh, bobp, getting a linker fix from Apple. Hill, markm the 
trouble with malloc. Bill, michael, NEWS versions. Dean, marcs, 
backup versus archive. Bobp, markm, should the simplified interface 
be in C or C++ this week? Bobp, marcs, resources for documentation 
again. 

Anyway, this week's task list is as follows:

A.       dean

x    6/12/90     shepherd recipe tested
x    6/12/90     bogus smalltalk snarf tested(1.5)
x    6/12/90     Transmitter compute sizes & pointers smalltalk version tested(1.5)
x    6/12/90     view shell tested(2)
x    6/12/90     Transmitter to copy data smalltalk version tested(1.25)
x    6/12/90     snarf packer, bogus allocation in view handler tested
x    6/19/90     ent saying changed and commit (8)
     6/19/90     move to new xanadu version
     6/19/90     make dagwoods snarf
     6/19/90     boot object
     6/19/90     receiver, smalltalk version tested(1.25)
     6/26/90     ent saying may commit and destroy view (5)

B.       hill

x    6/12/90     resolve malloc temporarily
     6/19/90     freeze alpha 6
     6/19/90     enhance translator with ixx
     6/26/90     find and kill garbage collection usage bug!
     6/30/89     write mail about X++ upgrades and garbage collection for bobp

C.       hugh

     6/19/90     resolve linker bug
     6/19/90     try client-only version of Xanadu for mac
     6/19/90     compile fm on mac (3)
     6/18/90     write event generator (2)
     6/18/90     doing image (2)
     6/25/90     implement warp (4)
     6/25/90     run compositors (2)

D.       markm

x    6/19/90     bootstrapping problem 
x    6/19/90     track endorsed ids and endorsing club ids (16)
     6/26/90     changing permissions prop by read clubs (22)
E.       marcs

x    6/12/90     run ian through gauntlet
x    6/12/90     review system overview
     6/19/90     get ian comments from michael, roger
     6/19/90     review old issues
     6/19/90     talk with chris hibbert at decwsl
     6/24/90     start second round of patenting work
     6/30/89     figure out whether links are standalone objects

F.       michael

x    6/12/90     understand dean's rough protocol
     6/19/90     bogus snarf handler, just memory
     6/19/90     copy data transceiver

G.       ravi

     6/12/90     D and L over febe in c++ (9)
     6/19/90     exhaustive D&L test (6)

H.       bobp

x    6/12/90     rough draft of simplified servet interface (SSI)
     6/19/90     polished draft of simplified xanadu interface (SXI)
     6/19/90     help hugh hit apple on the linker
     6/26/90     implement SXI

I.       roger

x    6/12/90     make inline work
     6/19/90     get fm running under open look
     6/19/90     xanatext in xwindows
     6/12/90     get r3 from ftp for rick

J.       roland

x    6/12/90     fix stubble for roger
     6/12/90     fix stubble for rick
     6/19/90     fix line numbering in crtuch
     6/19/90     run ceran over the whole world
     6/19/90     formic upgrade for recognizing subclasses of shepherds, for dean/michael
     6/26/90     fix stubble for 1-way connections for dean
     6/26/90     xlint look for missing redundant curlies
     6/30/90     write grammar for formic
     6/30/90     parser with expression semantics
     6/30/90     xlint check for assign to a parameter variable and assignment to "this"

K.       Bill

x    6/05/90     fix dead cpu, dead disk
     6/19/90     figure proxy level protocol
     6/19/90     figure out dependencies 

L.       Joel

x    5/15/90     distribute info packet to world
x    5/15/90     molecular negotiations

M.       Jacque

     6/19/90     more outline of tapestry documentation
     6/26/90     new flat view scroll bars

N.       Rick

     6/19/90     formic automated generator of interface
     6/19/90     hire carey
     6/19/90     models for manipulators

O.       Paul

x    6/12/90     decide how to fix namespaces and gluing
     6/12/90     make scheme programmable

L.       Small Tasks List
    Cleanup region code
    Sequence # in smalltalk, dean & michael versions
    Test files

L.       Deferred Meetings:

     Roger, hugh, reproducible Tapestry tests. 
     discuss stubble for ADI with autodesk
     Bobp, joel, marcs, discuss Apple and smalltalk. 
     Marcs, dean, joel, parc place strategy. 
     Dean, michael, structured debug on the ent. 
     Michael, dean, break up the ent for the disk on Tuesday. 
 
    Michael, dean, markm, ravi, efficiency of internsets.
     Roger,, hugh, dean, ravi discuss system independent attributes
          for the frontends  after fm does paragraphs.
     Roland, dean, discuss xlint commands in smalltalk. 
     Markm, hill, discuss explicit deletion of orgls & refcount deletion. 
     Markm, hill, discuss exception handling on sensors between comm handlers.
     Dean, roger, michael, markm, discuss sequence numbers of size 4 billion. 
     Dean, markm, discuss intersecting backfollow ideas. 
     joel, marcs, discuss terminology in X++ for developer release 
             (particularly necromancers and related tools)
     Ravi, roger, discuss common view.

M.       Unassigned

     merge contact databases
     scenario documents (markm, dean
     Tumblers as subclass of partiallyOrdered
     tutorial for users on the frontend
     reference manual for users on the frontend
     test scaffolding, code exercisers
     rollin, rollout a la Univac
     subaccounts, permissions on FE
     sys admin, i.e., archiving and partitioning disks

--marcs