Development of time shared basic system processor for Hewlett Packard 2100 series computers
by John Sidney Shema
A thesis submitted to the Graduate Faculty in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE in Electrical Engineering
Montana State University
© Copyright by John Sidney Shema (1974)
Abstract:
The subject of this thesis is the development of a low cost time share BASIC system which is capable
of providing useful computer services for both commercial and educational users.
The content of this thesis is summarized as follows:First, a review of the historical work leading to the
development of time share principles is presented. Second, software design considerations and related
restrictions imposed by hardware capabilities and their impact on system performance are discussed.
Third, 1000D BASIC operating system specifications are described by detailing user software
capabilities and system operator capabilities. Fourth, TSB system organization is explained in detail.
This involves presenting TSB system modules and describing communication between modules. A
detailed study is made of the TSB system tables and organization of the system and user discs. Fifth,
unique features of 1000D BASIC are described from a functional point of view. Sixth, a summary of
the material presented in the thesis is made. Seventh, the recommendation is made that error detection
and recovery techniques be pursued in order to improve system reliability. In presenting this thesis in partial fulfillment of
the requirements, for an advanced degree at Montana State
University, I agree that permission for extensive copying
of this thesis for scholarly purposes may be granted by my
major professor, or, in his absence, by the Director of
Libraries.
A program listing of the 1000D Time Share BASIC
system is not submitted as part of this thesis since pro-'
gram development was performed by a private business corp­
oration which considers the listing as proprietary infor­
mation.
Arrangements must be made with a representative of
Western Telecomputing Corporation of Bozeman, Montana in
Order to obtain a program listing for evaluation.
It is
understood that any copying or publication of this thesis
for financial gain shall not be allowed without my written
permission.
Signature , f y i r f w J d & b u t /
Date
Ti?
:
/T /9 7 ^ 7
DEVELOPMENT OF TIME SHARED BASIC SYSTEM PROCESSOR
FOR HEWLETT PACKARD 2100 SERIES COMPUTERS
by.
JOHN SIDNEY SHEMA
A thesis submitted to the Graduate Faculty in partial
fulfillment of the requirements for the degree
MASTER OF SCIENCE
in
Electrical Engineering
Approved:
Head, Major D e p a r # #
//
Chairman, Examining Committee
____ __________
Graduate Dean
MONTANA STATE UNIVERSITY
Bozeman, Montana
March, 1974
•
}
>6
iii
ACKNOWLEDGMENT
The development of any major computer software oper­
ating system is a lengthy, and complex process requiring the
cooperation and creative talents of many individuals.
The
author wishes to thank the programmers and system analysts
at Hewlett Packard Company for their excellence in designing
and documenting the HP 2000B time shared BASIC system which
served as a starting point for this investigation.
The writer is deeply grateful to Western Telecomputing
Corporation for supplying the facilities and equipment to
develop and debug 1000D BASIC.
He is especially thankful to
Dr. Donald K. Weaver Jr. for his backing of the project and
many suggestions for improvement.
J.S.S
iv
i
TABLE OF CONTENTS
Chapter
I.
Page
INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . ...
I
Historical Development of Time Share Principles. . . . .
i
Relevance of Previous Wor k. . . . . . . . . . . . . . . . . . . . . . . 6
Organization of Remaining Chapters. . . . . . . . . . . . . . .
7
II.
OPERATING SYSTEM REQUIREMENTS. . . . . . . . . . .
8
System Design Considerations. . . . . . . . . .
8
1000D System Specifications. . . . . . . . . . . . . . . . . . . . 11
III.
OPERATING SYSTEM SPECIFICATIONS. . . . . . . . . . . . . . . . . . . 16
User Software Capabilities. . . . . . . . . . . . . . . . . . . . . . 16
System Operator Software Capabilities. . . . . . . . . . . . . 22
IV.
TSB SYSTEM ORGANIZATION... . . . . . . . . .
27
TSB System Modules. . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
TSB System Tables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Disc Organization. . . . . . . . . . . .
54
V.
VI.
UNIQUE FEATURES OF 1000D. . . . . . . . . . . . . . . . . . . . .
59
ASSIGN Statement. . . . . . . . . . . . . . . . . . . . . . . . . . . .
CHAIN Statement. . . . . . . . . . . . .
FORMAT Command.. . . . . . . . . .
DISC Command. . . . . . . . . . . . . . . . .
59
67
73
78
SUMMARY AND RECOMMENDATIONS. . . . . . . . . . . . . . . . : . . . . 86
Summary. . . . . .
Recommendations. . . . . . . . . . . . . . . . . . . . . . . . . .
87
BIBLIOGRAPHY.... . . . . . . . . .
89
86
APPENDIX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Appendix A:
Appendix B:
Hardware Requirements.. . . . . . . . . . . . . . . 92
1000D Core Map./.. . . . . . . . . . . . . . . . . . . 96
V
LIST OF TABLES
Table
Page
1.
BASIC Matrix Operations. . . . . . . . . . . . . . . . . . . . . . . . . 17
2.
Base Page Links. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.
System Console States. . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.
System Library Overlay Programs.... . . . . . . . . . . . . . . . 42
5.
ID Code Entry Format. . . . . . . . . . . . . . . . . . . . . . . . . . . 45
6.
Base Page DIREC Format. . . . . . . . . . . . . . . . . . . . . . . . . 48
7.
DIREC Values. . . . . . . . . . .
8.
COMTABLE Format. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
9.
System Disc Organization. . . . . . . . .
10.
49
56
User/Library Disc Organization. . . . . . . . . . . . . . . . . . ...58
vi
LIST OF FIGURES
Figure
Page
1.
Terminal Data Formats. . . . . . . . . . . . . . . . . . . . . . . .
2.
Example of ROSTER Command. . . . . . . . . . . . . . . . . . . . . . . . 24
3.
Example of REPORT Command. . . . . . . . . . . . . . .
4.
Example of DIRECTORY Command. . . . . . . . . . . . . . . . . . . . . 25 .
5.
Bit-Flag Representation. . . . . . . . . . . . . . . . . . .
6.
I/O Buffer States.. . . . . . . . . . . . . . . . . . . .
7.
Disc Address W o r d . . . . . . . . . . . . . . . . . . . . .
8.
System Queue. . . . . . . . . . . . . . . . . . . . .
9.
Directory Entry Format. . . . . . . . . . . . . . . . . . . . . . . .
13
25
28
32
34
40
46
10.
Directory Pseudo-Entries. . . . . . . . . . . . . . . . .
47
11.
ADT Entry Format. . . . . . . . . . . . . . . . . . . . . .
12.
COMTABLE Entry Coding Format... . . . . . . . . . . . . . .
53
13.
HP7900 Disc Organization. . . . . . . . . . . . . . . . . . .
54
14.
HP7901 Disc Organization.... . . . . . . . . ... ■. . . . . . . . 55
15.
ASSIGN Syntax Processing Flow Diagram. . . . . . . . . . . . .
16.
ASSIGN Execution Flow Diagram. . . . . . . . . . . . . . . .
17.
CHAIN Statement Syntax Processor. . . . . . . . . . . . . . . . . .
69
18.
CHAIN Execution Flow Chart. . . . . . . . . . . . . . . . . . . . . . .
71
19.
FORMAT Command Flow Chart. . . . . . . . . . . . . . . . . .
20.
DISC-UP Command Flow Diagram... . . . . . . . . . . . . . . . . . . .
80
21.
DISC-DN Command Flow Diagram. . . . . . . . . . . . . . . . . . . . .
84
49
62
64
74
v ii
ABSTRACT
The subject of this thesis is the development of a low cost
time share BASIC system which is capable of providing useful com­
puter services for both commercial and educational users.
The content of this thesis is summarized as follows: First, a
review of the historical work leading to the development of time
share principles is presented. Second, software design considerations
and related restrictions imposed by hardware capabilities and their
impact on system performance are discussed. Third, 1000D BASIC
operating system specifications are described by detailing user soft­
ware capabilities and system operator capabilities. Fourth, TSB
system organization is explained in detail. This involves presenting
TSB system modules and describing communication between modules. A
detailed study is made of the TSB system tables and organization of
the system and user discs. Fifth, unique features of 1000D BASIC
are described from a functional point of view. Sixth, a summary of
the material presented in the thesis is made. Seventh, the recom­
mendation is made that error detection and recovery techniques be
pursued in order to improve system reliability.
CHAPTER I
INTRODUCTION
Time sharing is a technique of computer system software
design which allows many users, located at remote terminals, to
have simultaneous access to a single central computer.
Due to the
extremely fast processing speed of the computer in comparison with
that of the terminals, the computer is able to share its resources
among the users.
This makes it appear as if each terminal was
directly connected to its own separate computer.
Even if one ter­
minal is running a program requiring large amounts of computer time,
the other user terminals are required to wait only a few seconds for
the computer to process their programs.
The objective of this thesis is to present an overall des­
cription of one specific time share system.
This system was devel­
oped by Western Telecomputing Corporation in order to provide low
cost computer services for both commercial and educational organiz­
ations in Montana
HISTORICAL DEVELOPMENT OF TIME SHARE PRINCIPLES
The introduction of the minicomputer in the late 1960's
along with its associated low cost-high performance peripheral
2
equipment, caused the data processing industry to undergo a massive
change.
Early computer facilities emphasized the need to create
powerful computer centers whose power was measured in terms of
physical dimensions such as number of central processors, amount of
main core memory available, and number of peripherals attached.
Such
thinking lead to the creation of multi-mil I ion dollar computer systems
capable of handling tremendous quantities of problems and data.
However, the cost of such systems was prohibitive to smaller companies
and educational institutions.
With today's technologies of microminiaturization and large
scale integration of circuits having reduced the size and cost of
computers, a new philosophy of data processing has emerged.
Rather
than have one extremely powerful computer facility to handle all
problems, it is now more economical to have several small processors,
each handling a specific problem.
In fact, in some cases the best
solution to a specific problem is to design a special purpose pro- .
cessor using state of the art techniques rather than attempt to pro­
gram a general purpose computer to perform the same function.^
Historically, much data processing was accomplished by pro­
grammers writing programs to a particular customer's specifications,
^ Kampe, Thomas W., "The Design of a General-Purpose Microprogram^Controlled Computer, with Elementary Structure," in Computer
Structures: Readings and Examples, pp. 341-347.
3
entering the program with data cards into a batch processing
facility, and returning hours later to see if any error had occurred.
Omission of even a single comma could mean the run produced no
results.
Additionally, time was required to analyze the printout to
determine if the algorithm written had actually performed the desired
task.
If not, several more program runs, each printing diagnostic
information would be required before the program ran error free.
Such a procedure was acceptable after a program had been fully written
and made free of logic errors.
routine basis.
The program could then be run on a
However, this procedure is unacceptable to the pro­
grammer since little, if any, interaction with the program was allowed.
Another problem which was not handled by the large computing
system was that of the small or medium sized program requiring
only several seconds of computer processing time.
In most cases,
the person waited several hours for his program to be run on the
computer even though actual execution time was less than a minute for
the entire program.
In many cases, the computer's central processor
was idle, while it waited for the results of a previous computation
to be printed.
. Numerous attempts were made to solve the problem of obtaining
maximum utilization of a computer's resources in the early 1960's.
of the most effective solutions was that of a time share system
developed at. Dartmouth College by John G. Kemeney, Chairman of the
One
4
Department of Mathematics, and Thomas E. Kurtz, Director of Com­
putation Center.
Their idea, which was first proposed in 1962, was
to design a program language which would be both simple to learn yet
powerful enough to perform complicated tasks.
The language was
implemented in 1964 and was termed BASIC, for Beginner's All­
purpose Symbolic Instruction Code.
Up to 40 simultaneous users were
allowed to communicate with the BASIC Language processor directly
through keyboard-printer terminals.
A master program, called the
supervisory control or executive, kept track of all communication
between the users and computer.
It decided who was to run next, how
long they were allowed to run, and performed the appropriate task
called upon.
-
An important feature of the Dartmouth BASIC system was that
the user was not required to be in the proximity of the computer in
order to run a program.
Terminals could communicate directly with
the computer over regular telephone circuits.
Thus, computer service
was never farther away than the nearest telephone.
The major advantage of a time share computer system over a
batch system is the ability of the programmer to communicate directly
with a "virtual processor" and have each line of code examined
immediately upon entry into the computer.
This allows for on-line
debugging of programs which is not possible on most batch systems.
Since the computer is being "shared" among many customers, the cost
5
of computer services can be shared among the various.customers.
allows even small businesses to afford computer services.
This
Finally,
communication with a time-share computer is on a real-time basis.
Data can be entered and analyzed immediately, with results printed
on the user terminal in a matter of seconds rather than hours.
This
capability of time share gives a company the ability to analyze data
on sales, inventory, expenses, etc. and receive up to the minute
company reports, such as cash flow or sales forecasting.
The Dartmouth BASIC system, capable of supporting many terminals,
required a substantial investment in terms of processors, discs, and
peripheral equipment.
One of the first breakthroughs in terms of
providing a low cost but powerful BASIC Language processor was the
announcement by Hewlett Packard Company of their minicomputer-based
2Q00A Time Shared BASIC System in 1969.
The 2OOOA System could
generate up to 16 channels of BASIC for an investment of approximately
$100,000.
The 2000B System generated up to 32 channels of BASIC for
about the same initial investment.
The only major problem was that
the initial system had limited disc space and additional disc space
was very expensive. .
In 1972, Western Telecomputing Corporation started development
of a Hewlett-Packard 2000B - based time shared BASIC system designed
around a new low cost moving-head disc drive, the Hewlett Packard
7900A. Design specifications stated that the system should be capable .
6
of supporting up to 16 channels of BASIC, meet most of HP2000B Language
specifications, provide large quantities of disc storage, and have
a total system cost of less that $40,000.
The version of BASIC that
was developed to meet these requirements was called the WTC-I0000
Time Share BASIC System.'
RELEVANCE OF PREVIOUS WORK
The WTC-I000D time shared BASIC system described in this thesis,
is only one of the many variations upon the Dartmouth System as con- .
ceived. by Kemeney and Kurtz. A study of all manufacturers of computers
would show that each one has his own version of BASIC.
Some of these
are so powerful and have so many extensions that they tend to obscure
the simplicity of the original language.
The system described is organized almost identically to that of
Hewlett Packard 2000B., That particular implementation of BASIC was
selected for two reasons.
readily available.
The first is that the source tapes were
The second, and most important, was that the 2000B
system had been documented and described in sufficient detail that
modifications could be easily made and tested.
7
ORGANIZATION OF REMAINING CHAPTERS
Chapter 2 discusses restrictions imposed upon the design of
the system in terms of hardware arid performance requirements.
A
discussion of terminology is presented to acquaint the reader with
various concepts underlying the principle of time sharing.
Chapter 3 is a presentation of the operating system specifications
of 1000D BASIC.
It presents an overview of the system from both the
user's point of view and the system operator's point of view.
Chapter 4 describes TSB System organization.
It presents an
overview of TSB modules and describes communication between modules.
This chapter describes the organization of TSB tables and explains
their functions.
Finally, an account of disc organization is given,
both for system discs and user/library discs.
Chapter 5 is a description of unique capabilities of 1000D
which were developed to enhance system operation or make the system
better fit the needs of commercial customers.
The concluding chapter 6 gives a summary and conclusion of IOOOD
as a solution to providing economical computing services to both
commercial and educational users.
Finally, several recommendations
for system improvement are presented and discussed.
CHAPTER 2
OPERATING SYSTEM REQUIREMENTS
In designing a computer operating system, two types of
problems must be solved.^ The first type deals with intrinsic
problems or those which are inherent in the design itself.
The
designer must Took at the function the software is to serve and
evaluate the requirements needed to implement that function.
A
design consideration of 1000D BASIC was that it should perform a
useful function for both commercial users and educational users.
This decision immediately created several design problems.
Commercial customers needed large files for storage of data and
would be using programs requiring significant compute times.
Increased file access was needed and this would tend to tie the
system up.
Educational users demanded rapid response time to pro­
grams which would be mostly input/output bound.
data banks would not be required.
Access to large
Finally, a system of user
security would be required to prevent unauthorized access to a.
user's private files by another user.
The second type of problem occurs when one attempts to solve
the first, that is technological problems arise.
^ Walsh, Larry, 2000C Course Notes, p. 10.
These problems
9
can be looked upon as forming the constraints with which the system
designer is forced to work.
For example, limiting the amount of
computer core memory implies an upper limit to user program length.
Not all of the system coding may fit in memory so techniques must
be derived to move blocks of coding into core when they are to be
executed.
These constraints force the system designer to carefully
consider various algorithms for making maximum use of the system
resources.
If the designer has decided upon a system language, BASIC
in this case, communication with the system must be considered.
Two levels of communication are required . 3 One is a command lang­
uage.
A processor for this language is then required.
This
processor is responsible for handling system access, for control of
system resources, and for file and program control.
level of communication is that of a program language.
preter for that language is required.
The second
An inter­
This is the processor which
executes program statements that control file manipulations,
carries out calculations, and perform specific operations on data.
Other technological problems must be solved while planning
the operating system.
The hardware system must be separated by
software procedures into a number of virtual processors for
3
Ibid., p . 10.
10
simultaneous use by many users.
Most importantly, such sharing of
resources must be done in a way that makes sharing of resources
"invisible" to all of the users.
This places the requirement that
the system be highly organized.
Sharing of resources implies that a master program must
exist whose function is to determine which user will be allowed to
execute a command or perform a calculation.
referred to as a scheduler.
Such a program is
The time this program is running rather
than a user program is called overhead time.
Overhead is the time
the system spends performing routines designed to implement the
sharing of resources.
It is time lost to the users since their
programs are idle when.the scheduler is operating.
It is impor­
tant that the overhead time be kept to an absolute minimum in a
well organized system.
Another function of .the scheduler is to
maximize utilization of system resources. This means the
scheduler should attempt to keep each hardware subsystem busy.and
insure that the system does not "bottleneck" because one resource
is overly busy while others remain idle.
Response time or the amount of time required for the system
to respond to a user command will increase as system utilization
increases.
The system must wait for one user to complete a task
before it can allow a new user to start its own task.
The tech­
nique of "time-slicing" allows the computer to implement resource
11
sharing among the users and reduce response time.
This procedure
enables the. computer to set an upper limit on the amount of time
a user's program may execute in core.
If the program exceeds this
limit it is written back onto a disc from core and the next user's
program is read from the disc into core and allowed to execute.
IOQOD SYSTEM SPECIFICATIONS
WTC-I000D BASIC was designed to run on a Hewlett Packard
2100 series computer.
This section presents some of the design
specifications and constraints which were placed on I GOOD.
Appendix A contains a complete list of all hardware required to
operate the system.
1000D was designed as a small-scale time-share system capable
of supporting up to 16 simultaneous users.
In order to keep response
time of the system to less than five seconds, data communication
transfer rate was set to be in the 10 to 30 character per second
range.
Faster rates would force the scheduler to spend most of
its time character processing, leaving little time to execute user
programs.
The most common time-share terminal in use at the time of
system design was the ASR-33 teletype which operated at 10 characters
per second ( H O baud).
This terminal communicated in a bit-serial
12
A^ClI .code.
ASCII code was also the method of character represen­
tation internally within the HP 2100 series computers.
Hence, no
code conversion was required by the input/output processor.
Figure
I gives the format of the data words generated by a device operating
at 10 characters per second and by one operating at 30 characters
per second.
The input/output processing routine required at the worst
case .15 milliseconds to process a single character.
At an input
speed of 10 characters per second, the routine has two bit lengths
to process a character without loss of data, or 18 milliseconds.
It is obvious that at least 100 channels could be processed at this
speed.
For terminals operating at 30 characters per second, timing
is more critical.
The computer has only one bit length or 3.33
milliseconds to process a character from each channel without loss
of data.
Calculations show that the computer can process at best,
only 22 channels at 30 characters per second.
Therefore, a full
system of high speed terminals would tend to run much slower than
a full system of slow speed terminals since most of the computer's
time would be spent processing characters.
Since all communication from users to the system would be
over telephone connections, a second design consideration was to
allow system control of the telephone circuits.
reasons for doing so.
There were three
First, a person dialing a wrong number could
13
START
BIT
0
2
I
<---
3
4
6
5
S bit character code
---------------
STOP STOP
BIT BIT
7
--- > <18 msec >
IOO milliseconds------------------ >
Data Word at 10 Characters per Second.
START
BIT
0
I
2
3
4
5
6
7
d bit character code
S
33 .3 milliseconds Data Word at 30 Characters per Second.
Figure I.
Terminal Data Formats
STOP
BIT
3.33
msec
14
tie up one of the channels, preventing, others from accessing the
computer.
To solve this problem, the computer allows only a certain
number of seconds for the caller to log on the system.
to do so, the line is disconnected.
If he fails
Second, if the phones are on
a rotary call-in basis, a user can never be certain he will be able
to access a particular channel if he is accidentally disconnected.
A data set monitor in the system will log the user off the computer
in the event his telephone carrier signal is lost.
Third, a user's
telephone is automatically placed back on hook when he logs off,
making the channel immediately available to other users. .
The amount of user space available for programming is very
important to the customer and system designer.
like to have unlimited program space.
A customer would
The system designer is aware
that longer program areas mean the system will run slower since it
takes longer to swap users in and out of core.
A compromise was made
in 1000D setting maximum program length.to 4500 words, approximately
500 BASIC statements.
For longer programs, the user was given the
ability to "chain" programs together, resulting in virtually un­
limited program length.
A final design consideration was that of selecting an optimum
amount of disc storage.
Fixed head discs were ruled out since they
were very expensive and provided limited storage.
time was very fast, in the 5 millisecond range.
However, access
15
A decision was made to utilize the Hewlett Packard 7900 and
7901 moving-head disc drives.
Access time was in the 35 millisecond
range, fast for a moving head disc, and storage capacity of a single
disc was roughly four times that of a fixed head disc.
Each moving-
head disc channel provided 203 tracks of 6144 words per track for a
total channel capacity of over 1.2 million words.
This was enough
storage for approximately 960 medium-sized, programs.
1000D allows
up to four disc channels for a maximum capacity of over 4.8 million
words or approximately 3840 programs.
CHAPTER I I I
.OPERATING SYSTEM PROCEDURES
The language processor of IOOOD BASIC is the same as that .
for
HP2000B.4
This is an extended version of the original Dartmouth
BASIC as first described by Kemeney and Kurtz.^
This chapter
summarizes the main features of BASIC from both the user's and
operator's view.
USER SOFTWARE CAPABILITIES
Extended BASIC enables the user to perform complex matrix
operations' in a single statement.
Table I lists these operations
and defines their function.
The working size of a matrix is defined by the program through
the use of DIMENSION statements.
For example, DIM A (5,3), creates
a matrix of five rows and three columns.
These are maximum limits
which may be dynamically redefined in the program using the matrix
IDN, Z E R , CON, READ, or INPUT functions.
4
In the above example, MAT 4
Hewlett Packard Company, 2000B: A Guide to Time Shared BASIC,
205 p p .
5 Kemeney, John G., and Kurtz, Thomas E., BASIC Programming, 122 pp.
17
TABLE I.
Operation
Addition
BASIC MATRIX OPERATIONS.
Example
.MAT C=A+B
Description
Add matrix B to matrix A.
result in matrix C .
Put
Subtraction
MAT C=A-B
Subtract matrix B from Matrix
A. Result in matrix C .
Multiplication
MAT C=A* B
Multiply matrix A by matrix B.
Multiply matrix B by scalar A.
Result in matrix C .
Inversion
MAT C=INV(A) Generates matrix C as the in­
verse matrix of A.
MAT C=TRN(B) Generate C as the transpose ma­
trix of B.
Transposition
ZERO
MAT A=ZER
Set all elements of matrix A
equal to zero.
Identity
MAT A=IDN
Initialize
MAT B=CON
READ
MAT READ C
Input
MAT INPUT B
Print
MAT PRINT A
Generates A as the identity
matrix.
Sets all elements of matrix B
to one.
Reads.data from program DATA
statements into matrix C .
Reads in elements of matrix B
from user terminal.
Causes all elements of matrix
A to be printed on user ter­
minal .
Note: ,MAT READ may also input data from a user disc data
file. MAT PRINT may write data to a user disc data
file..
18
A = CON (2,2), redefines A to be a matrix of two rows and two columns.
Matrices can only be redefined as being smaller than that set up by
the DIM statement.
matrix is 2000.
The maximum number of elements allowed in a
An array of more than 2000 elements will produce
an error condition.
BASIC contains special variables and language elements for
manipulating string quantities.
than numeric values.
These are alphanumeric values rather
String variables are defined in the system by
declaring a single letter (A through Z) followed by the dollar sign
($).
The physical length of the string variable must be declared in
a dimension statement.
For example, DIM A$(36), sets up a string
variable A$ of 36 elements.
Maximum string length is 72 elements.
A
string value is a set of from I to 72 characters enclosed in quotes.
Most printing and non-printing characters of the ASCII code set are
valid string characters.
During execution of a program, the logical
length of the string is set.
This is the count of the actual number
of characters in the string.
For example, the statement, A$ = "SAMPLE
STRING", has a logical length of 13 characters even though the physical
length of A$ was specified as 36 elements.
Substrings may be referenced by the use of subscripted
variables.
In the above example, A$(2,6) gives the second through
sixth character or "AMPLE".
BASIC allows string, substrings, and
string variables to be used with relational operators.
They are com­
19
pared and ordered as entries are in a dictionary.
For example, the
following statements may be written:
100
IF A$ = "TEST"
THEN 400
200
IF A$(4,6) < B$(7,8)
THEN 500.
The first statement will cause the program to go to statement 400
only if string A$ has four elements which are exactly "TEST".
The second example will go to 500 if the fourth and fifth
elements of A$ are alphabeticalIy less than the seventh and eighth
elements of B$.
Thus, BASIC allows string values to be read, printed,
compared, and sorted with very little programming effort required.
Two special features of extended BASIC are the ability to
specify length of the terminal print line and ability to access.the
system clock.
Use of the WIDTH, command allows the user to inform the
system that the terminal may print only 20 columns of data per line
or up to 132 columns per line.
Use of the TIM(X) function allows
a program to obtain the current minute, hour, Julian day, and year
from the system real-time clock.
This information is useful in
identifying multiple runs of the same program or for reporting run
times of various programs.
A serious limitation of 1000D BASIC is its rather small user
program area of 4500. words.
Since some of this area is used for
program overhead, program lengths generally have a maximum length
of 3500 words.
This is sufficient to store a BASIC program of roughly
20
500 statements.
If arrays are dimensioned, the user loses two
words of program space for each element in the array.
Commercial
programs are typically very long and utilize large arrays for data
storage.
Use of the BASIC CHAIN statement allows-the programmer to
divide one large program into many smaller segments which are
stored on the disc.
As each segment of coding is needed, it is
called off the disc by the currently executing program.
Parameters
and other values may be passed from one program to the next through
COMMON or data files. . Use of the CHAIN statement gives BASIC the
capability of handling programs of virtually unlimited length.
All users have access to two levels of libraries.
first level is that of the system public library.
The
These are pro­
grams and files saved on the disc by the master user, A 0 0 0 . Users
may get these programs from the disc, list and punch them, modify
them, and execute them.
public library.
User A000 may protect any program in the
This makes them "run-only" to all other users.
Such a procedure would be used to prevent unauthorized copying of
proprietary software.
A user also has access to a second level of
libraries, an individual user library.
Programs and files stored
in an individual library are accessible only by the owner.
The most powerful program function of BASIC is its versatile
file handling capability.
Users may create data files on the disc
to allow for direct manipulation of large volumes of data.
Files
21
may be of two forms: sequential or random access.
Sequential files are treated as serial access storage devices.
A program starts reading or writing data as a serial list of data
items from the physical beginning of the file and continues on an
element by.element basis until either the physical end of file is
encountered or an end of data condition occurs.
During a serial file
write operation, the new data element is written immediately following
the previous data element.
To retrieve an item from the file, the
program must, start at the beginning of the file and. read through all
items until it comes to the desired data element.
Files may also be used in a random access mode.
This enables
the program to break the file into a series of logical subfiles
that can be modified independently of each other.
Each subfile is
a block of 128 words.
Data may be written into a file in either numeric form or
string variable form.
mixed in any file.
String and numeric data may be randomly
Use of the TYP(X) function enables the program
to identify what type of data the next file item is during a file
read. Numeric data requires two words of storage per number whereas
two string characters may be stored in a single word.
File lengths are variable from I to 48 sectors.
holds 128 words of data.
Each sector
Maximum file length is 6144 words.
program may reference up to 16 files at one time.
/
/
/
Each
This is equivalent
22
to having immediate access to 98,304 words of additional data storage.
Any program may access an unlimited number of files during execution
through the ASSIGN statement.
This statement allows a new file to
be referenced in the place of a previously defined file in the pro­
gram.
The system allows users simultaneous access of files.
The
first user program to reference a file is allowed read/write
privileges.
Al I other users accessing the same file are allowed
read-only access.
Users whose account identification codes begin
with an A (A000 through A999) always are allowed read/write access
to the same files.
This section has described some of the more important
capabilities of extended BASIC.
A complete description of user
capabilities may be found in the WTC-I000D BASIC user's guide.
SYSTEM' OPERATOR SOFTWARE CAPABILITIES
The system operator is responsible for maintaining the TSB
system, add.ing new accounts, modifying existing accounts, and re­
moving old accounts from the system.
These functions are accomplished
through the system operator's console using special system commands
and through the user account A 0 0 0 .
System control through user A000 is restricted to two functions.
23
Only programs stored under that account, become a part of the,
public library, accessible to all users.
Placing programs and files
in this library is the primary function of A000.
The second
function involves maintenance of a particular program in the public
library, the "HELLO" program.
This program is automatically run
on the user's terminal following a successful attempt to log on
TSB.
HELLO is a standard BASIC program whose function is to describe
the system, give the current date and time, and print important
messages from the system operator to the users.
Primary system control takes place through operator commands
entered on the system console.
One function performed is that of
limiting access of system resources by the users.
The system
operator controls resource allocation to up to 562 user accounts,
the maximum allowed by the system.
This is accomplished by setting
restrictions on each account through operator commands.
Each account
is assigned a unique ID code and password required for system
access.
The operator also specifies a maximum number of minutes of
connect time allowed each user and additionally, a maximum number of
sectors of disc storage the user is allowed.
The password, maximum
connect time and maximum disc storage allowed may be modified at
any time by the system operator.
If required, an entire account
may be removed from the system by an operator command.
The system operator may request TSB to generate reports
24
useful for evaluating system utilization.
The ROSTER command
generates a listing of currently active channels as shown in Figure
2.
ROSTER
. . . . . . . S231 .... AOOO AOOO
K341 . . . . . . . . . . . . . . . .
B260
__
C246
......
Figure 2. . Example of ROSTER Command
The first line lists, from left to right, the users logged
on terminals 0 through 7.
User S231 is on channel 2, user AOOO
is on channels 4 and 5, and user C246 is on channel 7.
line list users on channels 8 through 15.
User B260 is on channel 13.
The second
User K341 is on channel 8 .
Four dots indicate an inactive channel.
A listing of all accounts with the total connect time used
to date and disc space used to date may be obtained with the REPORT
command. This: command is used to generate a report for account
purposes.
An example of the REPORT command output is given in
Figure 3.
I
25
REPORT
ID
AOOO
C021
C240
5231
5300
TIME
00007
OOOOO
00666
05267
03406
DISC
02798
OOOOO
00082
00186
00292
Figure 3.
ID
AOOl
C060
C241
5232
UOlO
TIME
15339
00767
00311
00099
00311
DISC
02314
OOOlO
00027
00091
01241
ID
AOlO
Cl 40
C242
5239
Z623
TIME
01723
00583
00106
00043
20001
DISC
00003
00125
00088
00047
01024
Example of REPORT Command
ID specifies the account identification code of the user.
TIME is the total connect time used in minutes and DISC is the
amount of disc storage used in sectors.
The DIRECTORY command is used whenever the operator wishes to
obtain a listing of library programs and files.
The operator may
specify what type of directory is desired, a complete listing of all
programs and files on the discs or simply a directory for an individual
account.
Figure 4 gives an example of a directory for a single user.
DIR - 5232
NAME
*ALFA1
*ALFA2
*ALFA3
*ALPHA
*C0DE*
BA
MCDA
WORK
DATE
355/73
004/74
004/74
006/74
353/73
009/74
011/74
023/74
Figure 4.
SUB
0
0
0
0
0
0
0
0
TR/SEC
092/25
092/22
102/07
092/18
139/00
092/04
166/20
117/02
LEN
03
03
03
04 F
06
14 C
14
08 C
Example of DIRECTORY Command
26
The account identification code is listed under the column
ID.
The program or file name appears under NAME, the date the
entry was last referenced is given under DATE.
The disc channel
number is printed under SUB and the disc address of the entry in
the form track number/sector number is printed under the column
TR/SEC.
The length of the program or file in sectors is printed
under LEN.
If the entry is a file, an "F" is printed.
A "C"
prints if the program has been saved in a semi-compiled form.
"P" indicates the entry has been protected.
A
If none of the above
appears, the entry is a program.
Programs and files which have not been referenced since a
particular date may be removed from the system with the PURGE
command.
The disc space recovered is returned to the system.
The
system operator may also move programs and files from one disc to
another with the MOVE command.
This feature is useful when attempting
to recover programs from a defective disc. ,
Finally, the system operator has the ability to initiate an
orderly shutdown of the TSB system through the SLEEP command.
Back­
up discs may be written using utility programs after TSB has stopped.
These backup discs may be used to restore the system in the event of
a system failure.
CHAPTER IV
TSB SYSTEM ORGANIZATION
This chapter describes in detail the various modules which
comprise TSB and the communication which takes place between these
modules.
Appendix B is a core map of the system.
modules are distributed throughout core.
It shows how the
Since almost all communi­
cation is carried out in tables, a description of the format and
function of important TSB tables is given.
Finally, organization of
the system and user/library discs is discussed, since most TSB tables
are disc resident.
TSB SYSTEM MODULES
Time-share BASIC is written in several independent modules
which link together through core and disc resident tables and through
system linkage variables in core.
This section describes the various
system modules, defines their function, and lists linkage rules for
communication with other modules.
are given in Appendix B.
Core locations for these modules
Ten modules will be discussed:
(I) base
page, (2) power fail processor, (3) input/output buffers, (4) BASIC ;
language processor, (5) disc criver, (6) system console driver, (7)
multiplexor driver, (8) clock driver, (9) scheduler, and (10) system
■
library overlay programs.
28
Base Page. The first 1024 words of memory are directly
addressable by instructions located in any other part of memory.
For this reason, most system links are located on base page.
Memory
locations below IOOg contain interrupt linkages to input/output
processing routines and general temporary storage locations used by
various programs.
The system equipment table starts at address IOOg
and constitutes the core resident information about the system.
Immediately following the equipment table are system status variables
and input/output driver temporaries.
Some of these linkages are
bit flags, each bit in the word referring to a single port or channel
on the system.
Figure 5 gives an example of an arbitrary bit-flag
word.
15 14 13 12 11 IO 9 8 7 6 5 4 3 2 I O
I
"O" associated with channel 15
"I" associated with channel I'
Figure 5.
Bit-Flag Representation
From figure 5 it can be seen that a value of "0" or "I" may be
associated with a particular channel by setting the bit in the word
29
to the appropriate state.
A test of the state of the bit is used
to determine whether or not a specific condition exists on a partic­
ular port.
Table 2 lists base page links, defines whether they are
word or bit flags, and describes their function.
The remaining words on base page not contained in user area
are linkages to numerous subroutines located throughout memory and
general use constants shared by various routines.
The user swap area begins at approximately I300g. The first
four words are used to save run-time variables when the user is
swapped out of core to the disc.
These four locations store the
A-register, B-register, Extend bit, Overflow bit, and Program
counter when the user's program time-slice is exhausted.
The next
block of memory stores subroutine return addresses of the interrupted
program.
General usage information follows.
The first word of
available program space in the user's swap area is approximately
1774g.
Power fail processor. If the computer is running when a power
failure occurs, control passes to the power fail processor which saves
the current machine and input/output system status.
to indicate that this status has actually been saved.
A flag is set
The program
then halts the computer.
When power is restored, the flag is checked to determine whether
or not the power up process can be started.
When sufficient power
30
TABLE 2.
BASE PAGE LINKS
NAME
TYPE
COMMENTS
MAIN
Word
Points to word 0 of the teletype table of user
program currently in core. If none present,
MAIN = 0.
LIB
Word
Points to the location in the command table of
the disc address of the library routine in core.
LIB = 0 when none is present.
PLEX
Bit
Bit = I, port is full duplex.
UNABT
Bit
Bit = I, no abort allowed.
ABTRY
Bit
Bit = I, abort attempted.
ENDSK
Word
0 if disc not busy.
TIMER
Word
I when current program is timed.
MPCOM
Bit
Bit = I for multiplexor message to scheduler.
IOTOG
Bit
Bit = I for input; 0 = output.
TAPER
Bit
Bit = I for channel in TAPE mode.
CHNRG
Bit
Bit = I for CHAIN running.
HRLAG
Bit
Bit = I for $HELL0 running.
TERR
Bit
Bit = I if error occurred while reading tape.
CRLAG
Bit
Bit - I if program is in compiled mode.
PHM
Bit
Bit = I if carriage return received.
PHT
Bit
Phones are timing when bit = I.
PHR
Word
Number of seconds allowed to log on system.
0 = half duplex.
31
returns, the program restarts the computer using the saved system
status.
This insures that BASIC will be restored to the same opera­
ting state as existed prior to the power failure.
Input/Output buffers. Each channel on the time share system
has a 50 word (100 character) input/output buffer for communication
between the multiplexor and T S B . Each of these is a "circular"
buffer, having a physical start and end address, and a logical start
and end address.
The time share system always communicates directly
with these buffers.
Output operations merely place characters in the
appropriate slot in the buffer.
buffer.
Input retrieves a character from the
Figure 6 shows the three cases which are allowed to exist
within the I/O buffers.
In each of the three cases shown in Figure 6, the variable
BGIN is a fixed value pointing to the start address of the user's
physical buffer.
BEND is the name of a variable pointing to the
first character following the physical buffer.
value equal to BGIN+50.
It is a constant
While in input mode, BPNT points to the
character location into which the next character will be deposited.
During output, it points to the location of the character currently
being transmitted.
BSTR points to the first character of the most
recent buffer during input.
On output, it points to the location
into which the next character will be deposited.
during input.
buffer.
BHED is used only
It points to the first character in the first logical
32
BGIN
BEND
Current
Line
BSTR=BHEd
BPNT
(A) Channel in normal input mode.
Line 2
(cont.)
Line 3
Line I
t
t
BSTR
Line 2
I
BPNT
(B) Channel in TAPE mode.
BHED
User has been Queued.
Character being output
Current
Line
BPNT
BSTR
(C) Channel in output mode.
Figure 6.
I/O Buffer States
33
BASIC language processor. The BASIC language processor of
WTC-IOOOD TSB is essentially the same as that for HP2000B.
This is
a completely re-entrant program which is responsible for syntax
checking, program compilation and decompilation, and for execution of
BASIC programs. A complete BASIC processor description is beyond the
scope of this thesis.
Complete documentation of the HP2000B processor
may be obtained directly from Hewlett Packard Company.^
Disc driver. The TSB disc driver is responsible for all data
transfer to and from the system discs.
Any module of the system may
call the disc driver to perform a disc transfer.
are required for each disc transfer.
Three parameters
The first is the disc address
in the A-register given in standard format as shown in Figure 7.
The.disc address word is interpreted automatically by the
disc driver.
The driver determines how bits 15 and 14 are to be
decoded depending on which disc drive is being used with the system.
The second parameter passed to the disc driver is the core
address from which data is to be read or into which data is to be
written.
This parameter is passed in the B-register.
set to one for a read from disc to core.
Bit 15 is
Bit 15 is zero for a trans­
fer from core to disc.
Crandall, Ron, and McEvoy, Dennis, 2000B Time Shared BASIC Internal
Maintenance Specifications, pp. 171-226.
34
15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
v
D
D-field
\r
Cylinder
Sector
(HP7900 disc drive)
Bit 15 = drive number (0-1).
Bit 14 = 0 for cartridge disc within drive.
Bit 14 = I for fixed disc within drive.
D-field
(HP7901 disc drive)
Bits 15-14 = drive number (0-3).
Cylinder
(HP7900 and HP7901)
Bits 13-6 = track number (0-202).
Sector
(HP7900 and HP7901)
Bits 5-0 = sector number (0-47).
Figure 7«
Disc Address Word
35
The third parameter is the number of words of data to be trans­
ferred.
This value is given as a negative number in the base page
link entry WORD.
Upon initiation of a disc transfer, the variable ENDSK is set
to one.
It is cleared by the disc driver upon completion of data
transfer.
This value is tested by each module prior to performing a
disc transfer to insure that the disc is available for use.
System console driver. The operation of the system console
driver is determined by two flags located on base page, T35F1 and
T35F2.
Table 3 describes the operation of the console as signified
by these flags.
TABLE 3.
T35F1
STATE
T35F2
STATE
0
0
0
-I
-I
0
-I
-I
SYSTEM CONSOLE STATES
CONSOLE DRIVER
OPERATING MODE
Driver is accepting input (normal state).
I) input command received and being processed,
or 2) output terminated from a system command
which is to be reinitiated.
Normal output occuring.
Outputting, at the end of which the current
system command will be reinitiated.
36
The console driver is always in the input wait mode unless
processing a command or printing.
System modules may call this
driver by placing the number of characters to be printed in the Aregister with bit 15=0 if a carriage return/linefeed is to be appended.
The B-register is set to the core address of the output buffer.
If
bit 15=1 in the core address word, punching will occur in addition
to printing.
Multiplexor driver. The multiplexor driver is responsible for
handling data communication between the user terminals and the input/
output buffers. Communication with up to 16 terminals is handled by
the WTC-200B hardware multiplexor.
Complete programming specifications
may be found in the WTC-200B reference manual
?
Output to the multiplexor driver is performed on a character
by character basis using the output a character subroutine OUTCH.
The character to be output is placed in bits 6-0 of the A-register.
The address of word 0 of the user's teletype table is placed in the
B-register.
Data is then transferred using a JSB OUTCH5I instruction.
The OUTCH subroutine places characters into a user's buffer
until it is filled (99 characters), at which point the user is
suspended.
Since BASIC is fully re-entrant, this poses no problems 7
7 Western Telecomputing Corporations, Instruction Manual for the
WTC-200B Communication Multiplexor, pp.,48.
37
for that module.
Other modules of the system are not allowed to
suspend a user since they are not re-entrant.
These modules must wait
until the buffer is empty before they attempt to output data to any
channel. Furthermore, they must never let OUTCH suspend their
operation.
Most of these routines typically only partially fill the
user's output buffer then call a special suspension routine which
saves the current run-status.
Input from a user channel is allowed only when the user is in
idle or input.status, or while entering a program under TAPE control.
Upon completion of input (carriage return received) a test is made to
determine if the channel is in TAPE mode.
If it is, the appropriate
MPCOM bit is set to indicate to the scheduler that input is ready
for processing.
If not, a bit in PHM is set which causes the clock
driver to set the appropriate MPCOM bit 200 milliseconds later.
This
prevents the teletype from getting out of synchronism with the
multiplexor driver.
The scheduler will initiate the appropriate
action upon detecting the MPCOM bit as being set.
While a program is running and not printing data, the port
is set to the idle mode during which input is allowed to occur.
If
the "BREAK" key is pressed for at least 1/10 of a second, the multi­
plexor driver will set an abort request condition to stop the pro­
gram.
While a program is printing, the port is set to the full
duplex mode.
This allows data to be input even while data is being
38
output to that channel.
Pressing the "BREAK" key during output will
also set the abort condition.
The multiplexor driver recognizes when an abort attempt has
been requested.
abort is allowed.
It checks the port's status to determine if an
If the user is running a library program other
than Library or Catalog, the abort attempt is ignored since the
routine may be in the process of updating system tables.
At other
times when aborting could cause loss of system information, the port's
UNABT bit is set.
When an attempt to abort is made, the driver will
not set the stop condition but will set a bit in ABTRY. Routines which
set UNABT always call ABCHK when aborts will no longer cause any
problems.
ABCHK tests ABTRY and aborts the user if the channel's
bit was set.
Clock driver. The clock driver is the central point through
which control is passed to the scheduler.
Entry comes to the driver
whenever the clock (real time scalar) interrupts.
every 100 milliseconds.
This event occurs
Besides giving control to the scheduler,
the clock driver also performs the following tasks:
During normal
system operations, the console teletype motor is turned off.
Logging
operations and user messages automatically turn the motor on and set
up a two second delay to allow the motor time to reach operating
speed.
The clock driver is responsible for initiating output on
the system console after the two seconds.has elapsed.
39
When a user terminal is in the input mode and receives a
carriage return, the. multiplexor driver sets up a 200 millisecond
delay.
The clock driver tests each channel for this condition and
sets the port's MPCOM bit after the delay has timed out.
The clock driver also keeps the BASIC time of day clock.
This
is a real-time clock which keeps track of second of day, hour of day,
day of year, and century.
Finally, the clock gives control to the
scheduler.
Scheduler. The basic philosophy of the TSB scheduling algorithm
is to provide short response times for short, interactive jobs at
the possible cost of delays in longer running jobs.8
The implemen­
tation of this involves a queue of jobs to be run arranged according
to a priority scheme.
The queue is a list of from I to 18 entries,
each entry pointing to the next entry, and the last entry pointing
back to the first.
The 18 entries in the queue are the 16 LINK words
in the.user teletype table, a LINK word in the console teletype table,
and a queue head.
At the head of the queue are the locations MLINK,
MILNK+1, and MLINK+2.
These entries are always on the queue.
priority of the queue is stored in MLINK+2.
is the last entry in the queue.
8
The
It is always 77777g and
M L INK+I is the pointer to the first
Crandall, and McEvoy, op c i t ., pp. 17-19.
40
entry on the queue.
to itself.
In the case of an empty queue, MLINK+1 points
Each entry on the queue has a priority no greater in
value than that of the one it points to.
In adding a new entry to
the queue, the new entry is placed just ahead of the first entry
with a larger priority.
Figure 8 gives an example of a queue contain­
ing users one, four, and seven in addition to the system console.
Channel
number I.
TTY01+?LINK
?PLEV
Console
teletype.
T35LK
T35PR
Channel
number 4.
TTY04+7LINK
?PLEV
Channel
number 7.
TTY07+7LINK
?PLEV
Queue
head.
MLINK
MLINK+1
MLINK+2
Figure 8.
77777
System Queue
41
The scheduler uses the following rules to assign priorities:
1.
2.
Job priorities assigned upon initial placement in queue:
A.
SYNTAX and jobs returning from I/O suspend :0
B.
BASIC commands (RUN,LIST,PUNCH)
:1
C.
DISC Resident commands (GET,SAVE,etc.)
:2
Jobs are reassigned priorities as follows:
A.
Jobs of priority two are reassigned priority zero upon
reaching the top of the queue.
B.
RUN jobs which have exceeded their time sliced are
assigned priority four and repositioned in the queue.
Each job is assigned a time slice of one second.
System Library Overlay Programs. The last 256 words of non­
protected memory (37300g - 37677%) are reserved for execution of
system library overlay programs.
These are programs which are disc
resident and called into memory only when needed to perform a special
function, such as log a user on or off the system.
These programs
read into the overlay area and either run to completion or self
suspension during output.
Parameters required to restart the overlay
program are saved in the user's work area.
These programs use the
user swap area of 4500 words as a system routine work area.
are built and modified in this area by the overlay programs.
Tables
Table 4
lists the system overlay programs and describes their functions.
42
TABLE 4.
NAME
SYSTEM LIBRARY OVERLAY PROGRAMS
DESCRIPTION
FUSS Table
Table indicating files usage by the users.
FILES
Processes FILES statements in user programs.
CHAIN
Processes CHAIN statement syntax.
CHAIN.overlay
Executes CHAIN statement.
ASSIGN
Processes ASSIG statement syntax.
ASSIGN overlay
Executes ASSIGN statement.
SAVE
Stores a program in the user library.
CSAVE
Stores a program in semi-compiled form.
SUPERSAVE
Called by SAVE, CSA V E , and OPEN to redistredistribute
the disc directory on a full disc to gain disc
space.
GET
Retrieves a program from the library.
APPEND
Adds a program from the disc onto a program in
user area.
HELLO
Logs a user on the system.
BYE
Logs a user off the system.
KILL
Removes a program or file from the library.
RENUMBER
Reassigns the sequence numbers of a user program.
NAME
Gives a program a name.
CATALOG
Prints the names and lengths of all programs and
files in the user's library.
43
TABLE 4
NAME
(Continued)
DESCRIPTION
LIBRARY
Prints the names and length of all programs and
files in the public library.
WIDTH
Assign terminal print length from 20 to 132
columns.
DELETE
Deletes sections of a user's program.
TIME
Prints a listing of terminal time used since log
on, and total time used for the account.
DISC
Prints a listing of disc space used to date, and
total disc space allowed for the account.
MESSAGE
Sends a one line message from the user terminal
to the operator's console.
PROTECT
Used by AOOO to make programs read-only.
UNPROTECT
Allows full access to a previously protected
program. (A000 only).
OPEN
Creates a data file of the length specified by
the user.
LENGTH
Prints the length (in words) of the user's program
ECHO
Sets a terminal to full or half duplex.
REPORT
Prints a report of all user accounts, giving
total connect time and disc space used.
DIRECTORY
Prints a report of programs and files on the TSB
discs.
STATUS
Prints a report describing where TSB tables
are stored.
44
TABLE 4
NAME
(Continued)
DESCRIPTION
ROSTER
Produces a listing indicating active ports on
TSB.
DATjE
Changes the status of TSB's real time clock.
RESET
Resets terminal connect times to a specified
value.
CHANGE
Changes parameters in user ID table.
SLEEP
Initiates a systematic shutdown of the TSB
system.
NEWID
Adds an account to T S B .
KILLID
Removes an account from TSB.
MOVE
Moves a program or file from one disc subchannel
to another.
PURGE
Removes programs and files from TSB which haven't
been referenced since a specified date.
PHONES
Enables/disables the phone logic.
FORMAT
Formats a user disc.
COPY
Copies one disc subchannel to another disc sub­
channel .
DISC
Adds a new disc subchannel to TSB or removes a
disc subchannel from T S B .
ERROR TABLES
Six overlay programs containing error message
text.
45
TSB SYSTEM TABLES
System tables provide the cohesive framework that holds system
operation together.
This section describes the function and organ­
ization of five system tables: (I) user ID tables, (2) directory,
(3) disc availability table, (4) files usage table, and (5) command
table.
ID Table. The ID table (IDT) is a disc resident table which
contains one 8 -word entry for each user ID code in the system.
entries are kept sorted according to the ID codes.
the first code, Z999 the last.
AOOO would be
Each entry has the form shown in
Table 5.
TABLE 5.
WORD
0
1-3
ID CODE ENTRY FORMAT
CONTENTS
User ID code.
PASSWORD.
4
Time allowed in minutes.
5
Connect time used to date.
6
Disc allowed in sectors.
7
D i s c u se d t o
d ate .
The
46
The user ID is the internal representation of the user iden­
tification code such as AGIO.
Bits 14 through 10 identify the ID
letter ( A = I, B = 2, . . . , Z = 32g) and bits 9 through 0 identify the
ID number (0-999).
Password is a one to six character string.
are appended if less than six characters long.
is given in minutes.
Zeros
Time allowed and used
Disc allowed and used is given in sectors.
Words four through seven are 16-bit integers in the range 0 to 65535.
The following two words, located on base page, refer to the IDT:
IDLOC = disc address of IDT.
IDLEN = length in words of IDT.
Directory. The directory is a table which contains all necessary
information about each program or file in the system library.
are two directories on each disc.
There
A directory entry consists of
eight words and has the format shown in Figure 9.
Word
O
I
2
Contents
User ID Code
Program
Comments
Bit 15=1 if protected
Bit 15=1 if entry is a file
5
6
or File
Name
Bit 15=1 if semi-compiled
Start of program ptr. Starting core address
Date
Date entry last referenced
Disc address
7
-length in words
3
4
F i g u r e 9.
D i r e c t o r y E n t r y Format
47
The user ID is the system representation of the user ID code
in the same format as in IDT.
Words one through three are the name
of the program or file in ASCII characters, packed two characters
per word.
If fewer than six letters, spaces (40g) are appended.
The start of program pointer is used when "Getting" programs to tell
where in core to start reading in the program.
COMMON data to be passed between programs.
It is used to allow
Date is the day-of-year
and year the program or file was last accessed.
It is used by the
PURGE command to remove infrequently used programs and files from the
system.
The disc address in standard disc format and program length
are given for system access to the entry.
Entries are made in the directory alphabetically according to
words zero through three.
There are two pseudo entries in the
directory table to denote the beginning and end of the directory.
They have the form as shown in Figure 10.
O
I
2
3
4
O
O
O
O
O
O
I
2
177777
177777
177777
First
3
Entry
177777
O
5
6
177777
O
4
5
6
7
O
7
Fi g u r e 10.
Last
Entry
177777
O
O
Directory Pseudo-Entries
48
A core resident table called DIREC exists on base page.
contains information about the disc directories.
It
Its structure has
the form shown in Table 6.
TABLE 6.
BASE PAGE DIREC FORMAT
Word
0
1-4
Contents
-length in words of first directory track.
Same as first four words of first disc directory
track.
5
Unused.
6
Disc address of first directory track.
7-13
Same as
0-6 but applied to second directory track.
14-20
Same as
0-6 but applied to third directory track.
21-27
Same as
0-6 but applied to fourth directory track.
28-34
Same as
0-6 but applied to fifth directory track.
35-41
Same as
0-6 but applied to sixth directory track.
42-48
Same, as
track.
0-6 but applied to seventh directory .
49-55
Same as .0-6 but applied to eighth directory track.
y
49
A disc address of zero implies that there is no such directory
track.
When word 0 is zero, words 1-4 are meaningless.
DIREC
values are assigned values as given in Table 7.
TABLE 7.
DIREC VALUES
Contents
Word
0-13
Directory information for disc subchannel 0.
14-27
Directory information for disc subchannel I.
28-41
Directory information for disc subchannel 2.
42-55
Directory information for disc subchannel 3.
Disc Availability Table. The available disc table (ADT)
is a disc resident table which contains one two-word entry for
each area of the disc which is unallocated.
Figure 11 shows the
format of an ADT entry and the terminating pseudo entry.
O
Disc address
O
I-
Length of
Area
1
ADT Entry
Figure' 11.
177777
O
Terminating ADT Entry
ADT E n t r y Format
50
ADT entries are sorted according to word. zero.
Entries are
originally set one for each track available on the disc with word
s
one set to the maximum number of sectors on that track. As programs
.are stored on the discs, the ADT is searched for the first entry
containing sufficient disc storage space. After the program has
‘
been stored, the ADT entry sector count will be decreased by the
amount used.
A length of zero sectors indicates a track has been
filled and is no longer available for program storage.
As programs
and files are removed from the system, the released area is added,
back to the ADT.
Two ADT entries containing disc space released
in adjacent areas will be combined to form a single ADT entry.
Besides the entries for unallocated areas, there are ADT
entries for each of the two loader tracks, five TSB system tracks,
16 user tracks, IDT track, ADT track, and two disc directory tracks.
Word one of each of these entries is set to zero so they will not be
allocated for program storage.
In addition, during the loading
process, all defective tracks are removed from the ADT.
The ter­
minating ADT entry as shown, in Figure 11 denotes the end of the
table.
Since track zero is always allocated as a system track, any
possible disc address is guaranteed to be bounded by two ADJ entries.
The following two words on base page refer to the ADT:
ADLOC = disc address of ADT.
ADLEN = -length of ADT in words.
51
Files Usage Table. The files usage table (FUSS) is a disc
resident table used to control access of files by multiple users.
is divided vinto 16 subtables of 16 words per user.
It
Each subtable
lists the disc addresses of files currently being accessed by the
corresponding user.
FUSS is essential for preventing simultaneous
write access by two users of the same file, except for privileged
and semi-privileged users, and for preventing Killing a file when a
user has access to it.
capability.
The first user to access a file obtains write
All subsequent, but simultaneous, users get read-only
access except for users AOOO through A999 who always get read-write
access.
A user's FUSS table entries are set by the FILES and ASSIGN
routines.
These are both called from BASIC programs.
FUSS entries
are cleared by BYE, HELLO, KILLID, and sometimes by KILL.
The disc address of FUSS may be obtained by the instruction:
LDA
FUSS,I
Command Table. The command table (COMTABLE) is a list of all
user and system operator commands, containing their ASCII codings
and core or disc addresses.
Table 8.
The structure of COMTABLE is defined in
52
TABLE 8.
Section Label
COMTABLE FORMAT
Description
COMl
Codes for commands which are executed immediately
by the system.
COM2
Codes for commands which are executed by BASIC.
COM3
User commands which are executed by disc resident
programs.
COM4
System commands, all of which are executed by
disc resident programs..
C0M5
Core starting addresses for those commands which
are listed under COM! and COM2.
C0M6
Disc addresses for those commands listed under
COM3 and COM4. These addresses are filled in
by the TSB Loader.
Each command in COMTABLE is recognized by the subroutine SCOM
(search for command) only by its first three letters.
This search
routine takes the first three letters of any line input without a
statement number and converts each letter into a number from 0 to
31g, and then packs the three numbers into one word as three five
bit bytes.
An example of HELLO is given in Figure 12.
53
Letter:
ASCII
Offset
H
HO
101
E
L
105
101
114
101
7
4
13
L
X
E
L
7
4
15 14 13 12 11 10
O
I
X = ignored
difference
H
O
O
X
I
I
Figure 12.
I
6
9
O
8
O
7
I
2
13
6 $
4
3
2
1
0
O
O
I
O
I
I
O
I
3
COMTABLE Entry Coding Format
As shown in Figure 21, the COMTABLE entry for HELLO is 16213g.
In addition, all operator commands have bit 15 set to one.
This
indicates the command is valid only from the system operator console.
Some codes in COM2, COM3, and COM4 are set to minus one (all ones).
These do not correspond to any possible three letter code.
Their
function is to generate room in C0M5 and C0M6 for core and disc
addresses of routines called indirectly (such as FILES) or for tables
(such as FUSS). In the case of CTAPR, the purpose is to generate a
54
status type for printing compiler tape errors without a direct
command from the user.
DISC ORGANIZATION
1000D BASIC is designed to operate on either the HP7900A
disc drive or the HP7901A disc drive.
TSB can operate up to two
HP7900A drives or four HP7901A disc drives.
for the same amount of total disc space.
specified during system loading.
Either system provides
The disc drive type is
This section describes format of
the HP7900 disc system, format of the HP7901 disc drive, system disc
format, and user disc format.
HP7900 Disc Format. Two HP7900 disc drives may be used on TSB.
Each drive contains a fixed disc and a removable disc.
arranged into subchannels from zero to three.
Discs are
A minimum TSB system
would consist of subchannel zero which corresponds to the first re­
movable cartridge in disc drive one as shown in Figure 13.
Subchannel O
____ I_____
I
Drive I
(cartridge)
Subchannel I
(fixed)
Subchannel 2
(cartridge)
____ I____
I
Subchannel 3
(fixed)
Drive 2
Fi g u r e 13.
H P 7 9 0 0 Disc O r g a n i z a t i o n
55
HP7901 Disc Format. Four HP7901 disc drives may be used on
TSB, giving exactly the same configuration as exists on TSB with
HP7900 disc drives.
However, use of HP7901 drives allows greater
flexibility in disc handling since all discs are removable cartridges.
Figure 14 shows the organization of TSB using HP7901 drives.
Subchannel O
(cartridge)
Drive I
I
!Subchannel 2
(cartridge)
Drive 3
Subchannel I
(cartridge)
Drive 2
Figure 14.
I
Drive I
Subchannel 3
(cartridge)
HP7901 Disc Organization
Al I discs are directly interchangeable between 7900 and 7901
disc drives.
Discs in subchannels one through three are inter­
changeable and may be run on any other existing subchannel. However,
the disc in subchannel zero may only be run in subchannel zero since
TSB system and system tables are stored on that subchannel.
System Disc Organization. The disc in subchannel zero is
referred to as the system disc and must be present on a minimum TSB
system.
It contains a copy of the TSB system, user swap tracks.
56
and TSB system tables as shown in Table 9.
TABLE 9.
SYSTEM DISC ORGANIZATION
Subchannel 0
TRACK
SECTOR
0
0-2
TSB Disc Bootstrap Loader.
I
0-23
TSB LOADER.
2
0-47
CORE RESIDENT TSB (Part I).
3
0-47
CORE RESIDENT TSB (Part 2).
4
0-47
DISC RESIDENT TSB (Part I).
5
0-47
DISC RESIDENT TSB (Part 2).
6
0-47
DISC RESIDENT TSB (Part 3).
7
0-35
USER O SWAP TRACK.
8
0-35
USER I SWAP TRACK.
9
0-35
USER 2 SWAP TRACK.
10
0-35
USER 3 SWAP TRACK.
11
0-35
USER 4 SWAP TRACK.
12
0-35
USER 5 SWAP TRACK.
13
0-35
USER 6 SWAP TRACK.
14
0-35
USER 7 SWAP TRACK.
15
0-35
USER 8 SWAP TRACK.
16
0-35
USER 9 SWAP TRACK.
CONTENTS
57
TRACK
SECTOR
17
0-35
USER 10 SWAP TRACK.
18
0-35
USER 11 SWAP TRACK.
19
0-35
USER 12 SWAP TRACK.
20
0-35
USER 13 SWAP TRACK.
21
0-35
USER 14 SWAP TRACK.
22
0-35
USER 15 SWAP TRACK.
23
0-35
ID TABLE.
24
0-35
AVAILABLE DISC TABLE.
25
0-35
DIRECTORY TRACK I .
26
0-35
DIRECTORY TRACK 2.
0-47
USER AVAILABLE PROGRAM STORAGE.
27-202
CONTENTS
User/library Disc Organization.
Disc subchannels one through
three provide additional disc storage space for library programs
and files.
These discs are added to or removed from TSB through
the DISC-UP and DISC-DN commands respectively.
described in Table 10.
Subchannel format is
58
TABLE 10.
USER/LIBRARY DISC ORGANIZATION
Subchannels 1-3
TRACK
SECTOR
CONTENTS
O
O
DISC LABEL.
O
I
A D T , DIRECTORY I , DIRECTORY 2 LENGTHS.
O
2
SUBCHANNEL AVAILABLE DISC TABLE.
I
0-35
SUBCHANNEL DIRECTORY I.
2
0-35
SUBCHANNEL DIRECTORY 2.
3-202
0-47
USER AVAILABLE PROGRAM STORAGE.
CHAPTER V
. UNIQUE FEATURES OF IOOOD BASIC
WTC-IOOOD BASIC has four features which distinguish it from
HP2000B BASIC.
These are:
CHAIN statement, (3)
(I) ASSIGN statement, (2)
FORMAT command, and (4)
extended
DISC command.
This
chapter presents the details of these four features.
ASSIGN Statement
Extremely powerful files manipulation capabilities give BASIC
the ability to process large quantities of data.
the capability of accessing 16 files at once.
Each program has
However, for each file
named in a FILES statement, the user loses 128 words from his system
area.
Therefore, for relatively long programs, over 300 statements,
four or five files is generally the maximum number which can be
referenced.
Use of the ASSIGN statement allows a user program to reference
a virtually unlimited number of files.
It enables a program to assign
a new working file to replace a file set up by a FILES statement.
Statement format is:
S t a te m e m t #
ASSIGN
F ile
Name, F i l e
#,
R etu rn
V a ria b le
60
File Name is the name of a file given as a literal string of
up to six characters (seven if first character is "$") enclosed in
quotes or a string variable leading to a literal string.
' / in a string variable may not begin with
A file name
The "$" symbol as a
first character indicates a system file.
File number is a number of simple variable whose value is
between I and 16, indicating the position of the file to be replaced.
The return variable is one of the following five values which
is returned to the program when ASSIGN is executed.
The value returned
depends on the outcome of execution:
0 - The file is available for reading.and writing.
1 - The file is available on a read-only basis because it is
being accessed by another terminal. For users A000 through
A999, a return code of one indicates that the named file
is being accessed by another terminal. The file is still
available for reading and writing.
2 - The file is available on a read-only basis because it is
a system library file.
3 - The requested file does not exist or it is protected and
.the user attempting to "ASSIGN" it is not the owner.
4 - The file number in the ASSIGN statement is out of range.
It does not correspond to one of the positions reserved by
the "FILES" statement.
A flow chart of the ASSIGN statement syntax processor is given
in Figure 15.
The first step is to search the ASSIGN statement for a
61
file name.
If the name starts with a dollar sign ($), the reference
is to a public library file.
This sets a flag indicating that the
directory search routine is to perform a search using ID code AOO O .
After the name has been found, a test is made to determine if the name
is in the form of an alphabetic character followed by a dollar sign.
If not, the routine goes to the next section.
to a string variable is assumed.
table for the string variable.
Otherwise, a reference
ASSIGN syntax searches the symbol
If not found, an error return occurs.
If found, the name pointed to by the string variable is packed into
the file name buffer.
The second step in checking ASSIGN syntax is to search for a
file reference number.
or a simple variable.
file reference.number;
or a simple variable.
This may be either a one or two digit number
If a digit is found, the routine stores the
This may be either a one or two digit number
If a digit is found, the routine stores the.
file number and proceeds to the next section.
If a simple variable
is found, the symbol table is searched for the variable name.
error return occurs if an entry is not found.
An
Otherwise, the symbol
value is loaded, converted to an integer and saved.
If the value is
not defined, an. error condition occurs.
The final step is to search for a return variable.
Syntax
will only accept a simple variable which has an entry in the symbol
table.
Any other condition results in an error return.
62
Fi g u r e 15.
A S S I G N S y n t a x P r o c e s s i n g F l o w Diagram.
63
Fi g u r e 15.
(Continued)
64
^ File
number in
xX r a n g e /
B
DLOOK
Search
directory
for file
File
found
File
protected
Set return
variable
to 4
Return to
SCHBL
Set return
variable
Iy this
user
' Write
current
file to
disc
C
F i g u r e 16.
A S S I G N E x e c u t i o n F l o w Diagram.
65
F i g u r e 16.
(Continued)
66
Upon completion of syntax checking, another overlay program
is called to execute the ASSIGN statement.
Statement execution is
shown in Figure 16.
ASSIGN processing is responsible for replacing the old file
by the new one, setting the return variable to the correct value, and
updating the file table for the new file entry.
The first step it to
test if the file number references a file location given in a FILES
statement.
If out of range, the return variable is set equal to four
and the scheduler is called to set up execution of the next BASIC
statement.
Prior to returning, ASSIGN execution checks to see if the
user has attempted an abort.
If so, the user program is stopped.
If the file number is in range, DLOOK searches the disc direc­
tories for the entry given in the file name.
If no entry can be
found, the return variable is set to a value of three and a cell
is made to execute the next BASIC statement.
If an entry is found,
its type is checked to determine whether it is a program or a file.
If a program, control returns to the next BASIC statement with the
return variable set to three.
If the entry is a file, a test is made
to determine if it has been protected by a user other than that wishing
to ASSIGN it;. If protected, control returns to the next BASIC state­
ment with return variable set to three.
If the above tests show the
entry to be an unprotected file, the contents of the old file buffer
67
of 128 words are written to the disc.
The routine now tests for access to a file in the public
library by a user other than A O O O . If so, the value of the return
variable is set to two and a flag is set indicating the user is.
to be allowed read-only access.
The next step is to read in the files usage table (FUSS) and
clear the entry to the old file in the user's section.
A test is made
to determine if the new file is being accessed by another user.
If
it is, the value of return variable is set to one and the file is
flagged as read-only.
If the user is ID code AOOO through A999, the
return variable is set to one but the file is not flagged as read­
only.
The files usage table is updated, written back to the disc,
and parameters in the user's program area files table are set to
reference the new file.
Prior to.terminating ASSIGN execution, a final test is made to
determine if the file was flagged as read-only.
If not, the return
variable value is set to zero, otherwise its value remains unchanged.
CHAIN Statement
The CHAIN library routine processes, a CHAIN statement in the
68
user's program.
It consists of two disc resident routines, one which
checks syntax, the other performs CHAIN execution.
extended version of that in HP2000B.
1000D CHAIN is an
It allows the user program to
reference a program through a string variable and supply a starting
statement number for the CHAINED program.
Figure 17 is a flow chart
of CHAIN syntax.
CHAIN syntax processing is very similar to the first section'
of ASSIGN syntax processing.
The first step performed by CHAIN syntax
is to dump all current file buffers of 128 words each back to the disc.
It then searches for a program name.
Names starting with a dollar
sign set a flag indicating that a directory search is to be made
using ID code A000.
If the name consists of a letter followed by
a dollar sign, the symbol table is searched for that string variable.
If not found, an error message is printed.
Otherwise, the name
pointed to by the string is set as the program name.
A search is made to determine if a starting sequence number
has been supplied.
This parameter may be either an integer value or
a simple variable.
If it is given as a number, the value is stored.
If it is given as a simple variable, the symbol table is searched for
that entry.
If no match occurs, an error is printed.
Otherwise, the
value is loaded, converted to integer format and stored.
then calls the CHAIN execution overlay.
CHAIN execution.
Syntax
Figure 18 is a flow chart of
69
CHAIN
Load program
name
First
letter
/ Write
file
buffers
to disc ,
Error
return
/Null
program
name
Bad
name in
CHAIN
Set to
search
AOOO
Pack name
in buffer
Quote
found
String
operator
Operator
defined
Get name from
string and
put in buffer
F i g u r e 17.
C H A I N S t a t e m e n t Sy n t a x Processor.
I
70
Comma
follows
x. name
Search for
simple
variable
Variable
found
Value
defined
Load value
of simple
variable
Digit
Get statement
number
/ Bad
'name in
CHAIN
Valid
number
Error
return
Read
in
CHAIN
overlay
Figure I?.
(Continued)
71
DLOOK
Search
directory
, for
\ program
/ Read
I COMMON
area
SWAP arc
Read
user
program
in core
Found
ABCHK
Check for
aborts
File
Read
rogram
from
disc
SEMIC
Set up
pointers for,
\
Error
return
Too big
Update DATE
in
directory
/
ABCHK
Check for
aborts
Priority=A
Remove user
and insert in
QUEUE
Write
director
'
disc
BASIC
Move name
to user's
TTY table
j
Fi g u r e 18.
^Exit to
Scheduler
v (SCHl)
C H A I N E x e c u t i o n F l o w Chart
72
CHAIN execution searches the disc directories for the named
program.
An error occurs if the entry cannot be found.
entry type is checked.
message is printed.
Next, the
If a file has been referenced, an error
Otherwise, the entry is a program and a test is
made to determine if it will fit in core.
If it won't, an error
message is printed.
If the entry is a valid program which will fit into available
user space, the entry's DATE word in directory is updated and the pro­
gram's name is copied into the user's teletype table.
CHAIN execution then reads in the user's base page area, any
common area which existed from the previous program, and appends
the new program.
It then calls the subroutine SEMIC which sets up
pointers for BASIC describing the program as being either uncompiled
or semicompiled.
It also sets the program starting statement number,
to indicate where execution is to begin on the new program.
A check is made to determine if an abort attempt was made during
the last step.
If so, the user's program is stopped.
Otherwise, the
user's timeout clock is cleared, the user is taken off the queue, re­
inserted with a priority of four, and control returned to the scheduler.
73
FORMAT Command
Before a disc may be used on the TSB system, it must be placed
into proper user/library disc format.
Use of the FORMAT command
allows the system operator to format a disc for TSB operation.
This
command initializes the disc data fields, checks for bad tracks,
builds an available disc table for the subchannel and sets up directory
tracks.
Command format is:
FORMAT - < SUBCHANNEL NUMBER >, LABEL
Subchannel number is the disc subchannel to be formatted.
must be an integer value in the range one to three.
will cause an error message to be printed.
character label for the disc.
It
Any other number
Label is a one to six
A flow diagram for FORMAT is given in
Figure 19.
The first step in processing a FORMAT command is to scan the
console input buffer for a disc subchannel number in the range one to
three.
If not found, an error message is printed.
If a valid sub­
channel is read, the next step is to search for a one to six character
label.
Labels of less than six characters will have the remaining
label field filled with spaces.
Labels greater than six characters
will be truncated to only the first six characters.
74
FORMAT
Get
subchannel
number
Valid
subchanne
Print
error
message
Terminate
disc label
Set up
test pattern
Label
found
Print
error
message
Terminate
/ Read
/
in
FORMAT
overlay-
I
Fi g u r e 19.
F O R M A T C o m m a n d F l o w Chart
75
Test
upper 24
sectors
j
Bad tracks=0
Set disc
driver for
error return
Track
Error
occurred
/ Test
lower 24
sectors
Cannot
use thi
disc
Terminate
Error
occurred
Track is OK
Make ADT
entry
Index to
next
track
Track
Figure 19
(Continued)
76
Restore disc
driver to
normal mode
Write dummy
Directory
to
;
TR I SEC 0/
Write disc
label to
TR 0 SEC 0
Print
number of
bad tracks
Write ADT &
DIREC length!
to
j
TR 0 SEC I /
'
Terminate
Write ADT
to
TR 0 SEC 2
F i g u r e 19.
(Continued)
77
The second step is to build a test pattern of 3072 words to be
used in locating bad tracks.
3072 words is exactly 24 sectors or
one surface of a disc cartridge.
After the test pattern has been
built, the, FORMAT overlay program is called.
The first step of FORMAT overlay is to clear the bad tracks
counter and set, the disc driver to return upon detection of an
error condition.
The main loop initializes the data field of each
sector and reads and writes the test pattern on each disc surface.
This is a two step process, the top disc surface is checked first,
followed by testing of the lower surface of each track or block of
24 sectors.
If an error condition occurs, the data field of the bad
sector is marked as being defective and a test is made to determine
what track the error occurred on.
the disc may not be used.
If it is track zero, one, or two,
FORMAT prints an error message and aborts.
If the error occurs on a track other than zero, one, or two, FORMAT
skips placing the track number in the A D T . This prevents BASIC from
accessing bad tracks.
Tracks which are free from errors have their disc addresses
added to the subchannel available disc table.
Disc addresses are
base addresses; they contain no subchannel information.
After an
entry is placed in the A D T , the routine initializes and checks the
next disc track.
After all 203 disc tracks have been checked the.disc driver
78
is restored to its normal operating m o d e . The disc label is written
to track zero, sector zero in the form:
LBTSXXXXXX
wh e r e :
LB => DISC label follows.
TS => Time-Shared BASIC disc.
XXXXXX = One to six character disc label.
The negative lengths of the ADT and subchannel directories are
written to track zero, sector one.
sector two.
The ADT is written to track zero,
A dummy directory is written to track one, sector zero.
FORMAT terminates by printing the number of bad tracks found during
the initialization process.
The formatted disc may now be used on
I
the TSB system.
DISC Command
DISC subchannels one through three are used to provide additional
disc storage, for user/library programs and files.
An initial TSB
configuration consists of only subchannel zero, the system disc.
This
is the state of TSB following initial system generation, system update,
or restart from SLEEP.
Additional disc subchannels must be added to
79
TSB through the DISC-UP command.
Disc subchannels, except subchannel
zero, are removed from TSB using the DISC-DN command.
User/library
discs are interchangeable among subchannels one through three, and
between a TSB system operating with HP7900 disc drives and one with
HP7901 disc drives.
This section describes the operation of the
DISC-UP and DISC-DN commands.
DISC-UP Command. The.DISC-UP command is used by the system
operator to inform TSB that a new disc subchannel is to be added to
the system.
Command format is:
DISC-UP, <subchannel number>
Subchannel number is an integer in the range one to three,
referencing the disc to be added to the system.
DISC-UP is given in Figure 20.
A flow diagram of .
80
Get disc
subchannel
number
DISC-UP
Print
error
message
Valid x
subchannel
Terminate
Compute BASE
PAGE DIREC
for
subchannel
Disc
already
Read
disc
label
/ Print
Wd i s c on
/ system
Terminate
Read
disc
label
disc
Read in
DISC-UP
overlay
F i g u r e 20.
Read in
ADT &
DIREC
lengths
Terminate
DISC-UP Command Flow Diagram
81
Add subchannel
bits. Update
System ADT
Read in
disc
ADT
I
Set BASE
PAGE DIREC
for subchannel
Lead in disc
directory
/ --- f
one
Add
subchannel
bits
f
Add
subchannel
bits
\I
Write back
directory
two
<
Read in disc
directory
I
two
Print disc
on system.
Print disc
label
Fig u r e 20.
(Continued)
Write back
directory
one
Terminate
82
The first step in DISC-UP processing is to search for and
isolate the subchannel number.
Bad input or any value not in the
range one to three will cause an error message to be printed.
The
next step is to compute the location of the disc subchannel's base
page entry in DIREC. If the first word in BP DIREC is nonzero, the
disc is already up.
An error message along with the label of the
disc in that subchannel will be printed.
If the first word in BP
DIREC is zero, the disc label field is read.
The third and fourth
characters in the label field must be "TS", indicating it is a time
share disc.
If not, an error message is printed indicating the disc
is not a time share disc and may not be used.
If the disc is a time
share disc, the label is saved and the lengths of the disc subchannel
ADT and directory tracks are read.
The next step is to read the system ADT into core and find
where to insert the new subchannel ADT.
If an ADT overflow occurs,
entries from the new disc subchannel ADT are removed until the
combined ADT will fit in core.
Subchannel bits are added to the ADT
entries from the new disc and these entries are merged into the
system ADT at the correct location.
The updated system ADT is then
written back to the system disc.
After updating ADT, DISC-UP updates the base page DIREC
entries to point to the new disc subchannel. It then adds subchannel
bits to all entries on the subchannel's two directory tracks.
83
DISC-UP terminates by printing the subchannel number and its
disc label on the operator console.
DISC-PN Command. Disc subchannels are removed from TSB by
using the DISC-DN command. Command format is the same as DISC-UP.
A flow diagram of DISC-DN is given in Figure 21.
DISC-DN processing is essentially a reversal of DISC-UP
processing.
The first step is to isolate the disc subchannel and test
if it is already off the system.
If the subchannel has not been
removed from T S B , a test is made to determine if any user is currently
accessing files on that subchannel. If files are in use, an error
message is printed and the disc is not removed.
If no files are in
use, the system ADT is read into core and scanned for entries referring
to the disc to be brought down.
These entries are removed from the
system ADT and placed in a temporary A D T . The system ADT is written
back to the system disc.
Subchannel information is removed from all
entries in the temporary ADT and this table is written onto the
disc subchannel being brought down.
The two subchannel disc directories are processed by DISC-DN
to remove subchannel references.
Base page DIREC is cleared for the
downed disc and a message is printed on the system console to indicate
the subchannel has been removed from T S B .
84
Get disc
subchannel
/
C
(
\
DISC-DN
Print
error
7 message
- V Terminate
Compute BASE
PAGE DIREC
for
subchannel
Disc
already
down
Read
'in FUSS
table
Read in
System
ADT
Figure 21.
D I S C - D N C o m m a n d Flow Diagram.
85
Remove ADT
entries for
this disc
Write disc
DIREC one
back
Write
System ADT
to disc
Read in
disc DIREC
two
Remove
subchannel
bits from
disc ADT
Remove
subchannel
bits
Write
subchannel
ADT back
to disc
Write disc
DIREC two
back
Read in
disc DIREC
one
Print disc
removed from
system
j
Remove
subchannel
bits
Terminate
F i g u r e 21.
(Continued)
CHAPTER V I
SUMMARY AND RECOMMENDATIONS
. SUMMARY
This thesis has described IOOOD BASIC as one solution in an
attempt to provide low cost computer services for both commercial
arid educational users.
Chapter I presented background material on
the historic development of time sharing techniques.
It described
why Western Telecomputing Corporation initiated the project of
developing its own time share BASIC system.
Chapter 2 defined many of the terms used in describing time
share operation.
It presented a description of essential.operating
system requirements, placing emphasis on hardware capabilities and
the restrictions which they imposed on TSB design.
Chapter 3 defined TSB from two operating viewpoints.
first part described TSB from the user's viewpoint.
The
It described
how the system software capabilities appear to the user.
The
second part was a TSB description from the operator's point of view.
Chapter 4 was a detailed presentation of TSB system modules
and overall system organization.
major TSB modules.
The first part outlined each of the
It defined their purpose, structure, and
87
communication with other modules.
The second part gave specific
information about TSB system table organization.
Structure of the
tables was said to be the defining framework around which all modules
of TSB were written.
Finally, the chapter discussed how the system
and tables were organized on the discs.
.
BASIC.
Chapter 5 discussed in detail four unique features of IOOOD
These were the ASSIGN statement, extended CHAIN statement,
FORMAT command, and DISC command.
Each of these features were added
to TSB either to enhance system operation or provide useful additions
to the BASIC language processor.
RECOMMENDATIONS
Although WTC-I000D BASIC meets all the design specifications
and performance requirements as outlined in the previous chapters,
one design consideration has been purposely avoided.
This deals with
the problems of error detection and correction by the system.
This
is a problem which warrants further study.
In the design of I000D, it was assumed that if the hardware
was functioning properly, the software should be designed in such
a way that no errors would ever occur.
system which can have no errors.
This implies a software
Unfortunately, the time share
software is an extremely complex program and its operation is
entirely random, dependent upon the operations taking place at each
88
of the 16 terminals.
Thus, the programmer cannot guarantee that
the program is capable of handling the infinite number of special
cases which might occur at any time during normal operation.
Furthermore, BASIC has been written so that every instruction
must be correctly decoded and executed by the computer and that all
data transfer must take place without loss of even a single bit.
Only very slight attempts at error correction have been made.
For
example, if the disc drive recognizes it has made a data error,
it signals the computer and an attempt will be. made to perform the
transfer an additional ten times.
BASIC stops.
If it fails all these transfers,
No provisions are made to lock out the bad data track
after system initialization.
Thus, to insure system reliability the following provisions
should be added to BASIC.
an error.
First, attempt to establish what constitutes
Upon doing so, set up a system of error severity levels.
This should be based upon the consequences each error may incur
upon the system.
Next, devise a scheme to attempt an orderly
recovery from the error.
For example, if an attempt to store data
on a section of the disc fails, flag that section as being defective
and attempt to write to another section of the disc.
Finally, an
algorithm should be devised so that if. an error occurs which causes
system operation to stop, a recovery of the system is possible
without loss of previous data.
BIBLIOGRAPHY
90
BIBLIOGRAPHY
Crandall, R o n , and McEvoy, Dennis, 2000B Time Shared BASIC Internal
Maintenance Specifications, Hewlett Packard Company, Cupertino
Division, California, 1971, 226 pp.
Instruction Manual for the MTC-200 Communication Multiplexor, Western
Telecomputing Corporation, Bozeman, Montana, 1973, 48 pp.
Kampe, Thomas W., "The Design of a General Purpose Microprogram Controlled Computer with Elementary Structure", in Computer
Structures: Readings and Examples, pp. 341-347, compiled by
C. Gordon Bell and Allen Newell, McGraw-Hill, New York, 1971.
Kemeney, John. G., and Kurtz, Thomas E., BASIC Programming, John Wiley
& Sons, Inc., New York, 1968, 122 pp.
I POOP:
Operator's Guide, Western Telecomputing Corporation, Bozeman,
Montana, 1974
I POOD:
User's Guide, Western Telecomputing Corporation, Bozeman, MT
2000B:
A Guide to Time Shared BASIC, Hewlett Packard Company, Software
Publications, Cupertino Division, California, 1970.
2000B:
Operator's Guide to Time Shared BASIC, Hewlett Packard Company,
Software Publications, Cupertino Division, California, 1970
Walsh, Larry, 2000C Course Notes, Hewlett Packard Company, Cupertino
Division, California, 1971, 420 pp.
APPENDIX
APPENDIX A
HARDWARE REQUIREMENTS
IOOOD BASIC system consists of a Hewlett Packard 2100 series
computer, a moving head disc mass storage device, an operator's
console, a paper tape reader, a user input/output processor, and a
supervisory software program.
Figure 22 shows the logical organi­
zation of the system.
user
terminal
operator
console
up to 16 user
terminals or data
sets
user
terminal
Figure 22.
HP 2100
Series
Computer
Disc
Storage
user I/O
Interface
Paper tape
reader
WTC-1000D System Configuration.
.93
The system computer controls the allocation of resources
amont the users.
It controls the execution of user programs,
operations of the mass storage devices, communication between the
system operator and T S B , and communication between the system and
users through the I/O processor.
Table 11 lists the hardware re­
quirements for the system computer.
Table 12 lists the computer
input/output configuration.
TABLE 11.
HP2100 Series Computer Requirements
1.
16K words of main memory.
2.
2 channels of direct memory access (DMA).
3.
Extended arithmetic unit (EAU).
4.
Power fail/Auto restart.
5.
Memory parity check.
6 . Computer controlled clock (Time-base generator).
7.
Paper tape reader and interface.
8 . Moving-head disc memory.
9.
a.
HP7900A
b.
HP7901A
Operator console and interface.
10.
User I/O processor and interface.
11.
Up to 16 user terminals or data sets.
94
TABLE 12.
Hardware Configuration
I/O CHANNEL
CONTENTS
Disc Interface
IO-Il8
WTC-2OOB Multiplexor.
12S
IS8
Computer Controlled Clock
Paper Tape Reader
. 14S
Operator's Console
15S
The moving-head disc storage device is used for storage of the
TSB system, public library, and user files and programs.
drive is required for the basic configuration.
One disc
Optionally, one
additional HP7900 or three HP7901 disc drives may be added, to the
System. .
The operator communicates with the TSB system through the
operator console.
From this console the operator is able to start
T S B , add new accounts to the systemi, modify existing accounts, and
remove accounts from T S B . The operator may use the console to
generate reports on TSB system utilization, add or remove disc storage,
and systematically shut down the system.
The system console is also
used to keep a running listing of all user logging activity.
The main communication between the users and TSB is handled
over the user I/O interface system.
User communication is processed
95
by the WTC-200B Communications Multiplexor.
This device is capable
of processing up to 16 simultaneous user terminals.
The paper tape reader is used to load the initial system or
system update tape into the computer by the TSB loader program.
It is also used to read in the TSB bootstrap loader during system
restart.
APPENDIX B
:
IOOOD BASIC CORE MAP
BEGINNING OCTAL
CORE LOCATION
CONTENTS
O
23
100
243
272
A, B registers and I/O interrupt locations
System temporary storage
Core resident information about disc tables
System status variables
Multiplexor driver temporaries
307
435
622
. System linkage variables
General use constants
Language processor constants and addresses
1152
' Error jump table
1305
1311
1605
1626
. Beginning of user swap area
Subroutine return addresses
General usage
Syntax, compile, and execution information
temporaries
12112
user input/output buffers
13324
14000
Power fail/restart processor
Statement syntac processors '
14066
LET statement syntax
14111
14112
DIM statement syntax
COM statement syntax
14133
14173
■ DEF statement syntax
FILES statement syntax
14173
14173
14174
14200
CHAIN statement syntax
■ ASSIGN statement syntax
REM statement syntax
GOTO and GOSUB statement syntax
97
BEGINNING OCTAL
CORE LOCATION
H2L7
14317
14356
14362
14366
IF statement syntax
FOR statement syntax
NEXT statement syntax
RETURN, END, and STOP statement syntax
DATA statement syntax .
14430
. READ statement syntax
14440
14452
INPUT statement syntax
. PRINT statement syntax
MAT statement syntax
Formula syntax checker
14557
15000
15436
16553
20614
22000
22341
23676
24322
24641
25252
26072
27432
30160
31362
' 32011
32257
32502
32720
33417
34157
CONTENTS
•Syntax subroutines
Compilation and decompilation
Main execution loop and execution processors
Execution utility programs
Formula evaluation routines
- Execution utility routines
Arithmetic subroutines
Error routines
List routines
Matrix execution
Output routines
Library functions (SIN, COS, etc.)
ASCII to binary number conversion
Print name table
7900/7901 DISC driver
System console driver
Multiplexor driver
Teletype tables (Queue)
Clock driver
98
BEGINNING OCTAL
CODE LOCATION.
34254
Scheduler
35566
35732
Output character routine
Command table
36263
Library subroutines
Overlay program area
37300
CONTENTS
MONTANA STATE UNIVERSITY LIBRARIES
CO
milill HilIII 111
762 10015/174 7
I
N378
Sh 1+1+5
cop.2
Shema, John Sidney
Development of time
shared basic system
processor for Hewlett
_______Packard 2100 series..
*IAMC AIMD AOPWKa«
u2
-J,u, -dh'A
II2.i
y*\iar s
/6- "
SIi ?,
c(ien
«11(Q 'QL/u
/fii**
-
j“
^
A'37?