Comments:
Generally looks Good

1)DBC is PI not GCF
2)Discuss other peoples work:
  So we leverage libraries CHAOS DAGH Global Array
  Leverage HPF studies
  We participate in Java Grande activity and build lessons in
  I wonder if there is more directly related ideas -- Do you know
  what FORTRAN P (Woodward Minnesota)and F-- (Numrich Cray) are like
3)Somewhere discuss metacomputing i.e. targeting heterogeneous systems
  is a straightforward extension -- do this when we discuss thread implementation
4) Add a one page Project summary which has some technical depth 
  i.e. more technical than C.1

Now comments tied to particular pages:
5)Page 1,2 section C.1: say what we have achieved (i.e. adlib plus HPJava design and initial prototype
6)Page 1 End of paragraph 1: Perhaps remark that if EXPERT programmers build parallel programs and 
less experienced use rather develop, then not so suprisingly that SPMD popular
7)Page 2 First paragraph: Discuss regular v irregular briefly
8)Page 3, Goal 2: Add regular and irregular remark
9)Page 3, add Goal 3: Testing and Evaluating HPjava and HPspmd on large scale applications
10)Page 4, line 11: "learning naive use" seems poor english
11)Page 5, line 16: Add a short section summarizing enough of HPJava for rest of this section 
e.g. define overall used later
12)Page 5, line 8 from bottom: Somewhere here comment that here is where compiler optimizations come in
13)Page 6, line 12: Move sentence at end of section (Gosling in parentheses)here. In Goling
sentence change "the approved" to "his preferred"
14)Page 6, line 15: C becomes C and Fortran    and unrealistic becomes an unrealistic
15)Page 6, sec. C2.2 line 4: at Syracuse becomes led by Syracuse
16)Page 7, line 6: anticipate becomes believe
17)Page 9, line 2 from bottom: toolkit for becomes toolkit for several projects including
18)Page 11, line 20: Fast Multipole becomes Fast Multipole and its associated irregular MPI-based library
19)Page 12, HPspmd Collaboration model: don't go into so much detail -- just specify mutual visits
and continuation of ongoing electronic collaboration. Note substantial sharing of software already
as illustrated by HPF front end and f2j
20)Page 12 SUN work: Say this will look at Java for NCSA Alliance Grand Challenges
21)Around Page 12: Describe "application oriented for test and evaluation" development process
22)Page 15 line 18: Fortran should be Fortran90 ?
23)Page 15 bottom line: after HPF say (one in Europe and one at Syracuse)
24)Page 16: Comment on raw performance of adlib wrt PGI library

25)Section H
Say we have 8-CPU SGI Power Challenge , a cluster of about 14 Sun UltraSparcs used for a
mix of database, educational and computational resources, a cluster of 10 PC NT's
ATM already exists; SP2 and DEC machines have gone

26)I will get you a better summary of interests for me