minnie.tuhs.org/unixen.tar/unixen/PDP-11/Trees/V6/usr/doc/beg/u4

Text File | 1975-06-26 | 6.9 KB | 298 lines

.SH
IV.  PROGRAMMING
.PP
.UC UNIX
is a marvelously pleasant and productive system for
writing programs;
productivity seems to be an order of magnitude higher
than on other interactive systems.
.PP
There will be no attempt made to teach any of
the programming languages available on
.UC UNIX ,
but a few words of advice are in order.
First,
.UC UNIX
is written in C,
as is most of the applications code.
If you are undertaking anything substantial,
C is the only reasonable choice.
More on that in a moment.
But remember that there are quite a few programs already written,
some of which have substantial power.
.PP
The editor can be made to do things that would normally
require special programs on other systems.
For example, to list the first and last lines of each of a
set of files, say a book,
you could laboriously type
.B1
ed
e chap1.1
1p
$p
e chap1.2
1p
$p
 etc.
.B2
But instead you can do the job once and for all.
Type
.B1
ls chap* >temp
.B2
to get the list of filenames into a file.
Then edit this file to make the necessary
series of editing commands
(using the global commands of
.C ed ),
and write it into ``script''.
Now the command
.B1
ed <script
.B2
will produce
the same output as the laborious hand typing.
.PP
The pipe mechanism lets you fabricate quite complicated operations
out of spare parts already built.
For example, the first draft of the
.C spell
program was (roughly)
.B1
.ne 7
cat ...     (collect the files)
| tr ...    (put each word on a new line,
      delete punctuation, etc.)
| sort    (into dictionary order)
| uniq    (strip out duplicates)
| comm    (list words found in text but
      not in dictionary)
.B2
.SH
Programming the Shell
.PP
An option often overlooked by newcomers
is that the shell is itself a programming language,
and since
.UC UNIX
already has a host of building-block programs,
you can sometimes avoid writing a special purpose program
merely by piecing together some of the building blocks
with shell command files.
.PP
As an unlikely example,
suppose you want to count
the number of users on the machine every hour.
You could type
.B1
.ne 2
date
who | wc -l
.B2
every hour, and write down the numbers, but that is rather primitive.
The next step is probably to say
.B1
(date; who | wc -l) >>users
.B2
which uses ``>>'' to
.ul
append
to the end of the file ``users''.
(We haven't mentioned ``>>'' before _
it's another service of the shell.)
Now all you have to do is to put a loop around this,
and ensure that it's done every hour.
Thus, place the following commands into a file, say ``count'':
.B1
.ne 4
: loop
(date; who | wc -l) >>users
sleep  3600
goto loop
.B2
The command
.C :
is followed by a space and a label,
which you can then
.C goto .
Notice that it's quite legal to branch backwards.
Now if you issue the command
.B1
sh  count  &
.B2
the users will be counted every hour,
and you can go on with other things.
(You will have to use
.C kill
to stop counting.)
.PP
If you would like ``every hour'' to be a parameter,
you can arrange for that too:
.B1
.ne 4
: loop
(date; who | wc - l) >>users
sleep  $1
goto loop
.B2
``$1'' means the first argument when this procedure is invoked.
If you say
.B1
sh count 60
.B2
it will count every minute.
A shell program can have up to nine arguments,
``$1'' through ``$9''.
.PP
The other aspect of programming is conditional testing.
The
.C if
command
can test conditions and execute commands accordingly.
As a simple example, suppose you want to add to your login sequence
something to print your mail if you have some.
Thus, knowing that mail is stored in a file called
`mailbox', you could say
.B1
if -r mailbox  mail
.B2
This says
``if the file `mailbox' is readable, execute the
.C mail
command.''
.PP
As another example, you could arrange that the
``count''
procedure count every hour by default,
but allow an optional argument to specify a different time.
Simply replace the ``sleep $1'' line by
.B1
.ne 2
if $1x = x sleep 3600
if $1x != x sleep $1
.B2
The construction
.B1
if $1x = x
.B2
tests whether ``$1'', the first argument, was present or absent.
.PP
More complicated conditions can be tested:
you can find out the status of an executed command,
and you can combine conditions with `and', `or',
`not' and parentheses _
see
.SE if (I).
You should also read
.SE shift (I)
which describes how to manipulate arguments to shell command files.
.SH
Programming in C
.PP
As we said, C is the language of choice:
everything in
.UC UNIX
is tuned to it.
It is also a remarkably easy language to use
once you get started.
Sections II and III of the manual
describe the system interfaces, that is,
how you do I/O
and similar functions.
.PP
You can write quite significant C programs
with the level of I/O and system interface described in
.ul
Programming in C: A Tutorial,
if you use existing programs and pipes to help.
For example, rather than learning how to open and close files
you can (at least temporarily) write a program that reads
from its standard input,
and use
.C cat
to concatentate several files into it.
This may not be adequate for the long run,
but for the early stages it's just right.
.PP
There are a number of supporting programs that go with C.
The C debugger,
.C cdb ,
is marginally useful for digging through the dead bodies
of C programs.
.C db ,
the assembly language debugger, is actually more useful most
of the time,
but you have to know more about the machine
and system to use it well.
The most effective debugging tool is still
careful thought, coupled with judiciously placed
print statements.
.PP
You can instrument C programs and thus find out
where they spend their time and what parts are worth optimising.
Compile the routines with the ``-p'' option;
after the test run use
.C prof
to print an execution profile.
The command
.C time
will give you the gross run-time statistics
of a program, but it's not super accurate or reproducible.
.PP
C programs that don't depend too much on special features of
.UC UNIX
can be moved to the Honeywell 6070
and
.UC IBM
370 systems
with modest effort.
Read
.ul 3
The
.UC GCOS
C Library
by M. E. Lesk and B. A. Barres for details.
.SH
Miscellany
.PP
If you
.ul
have
to use Fortran,
you might consider
.C ratfor ,
which gives you the decent control structures
and free-form input that characterize C,
yet lets you write code that
is still portable to other environments.
Bear in mind that
.UC UNIX
Fortran
tends to produce large and relatively slow-running
programs.
Furthermore, supporting software like
.C db ,
.C prof ,
etc., are all virtually useless with Fortran programs.
.PP
If you want to use assembly language
(all heavens forfend!),
try the implementation language
.UC LIL,
which gives you many of the advantages of a high-level language,
like decent control flow structures,
but still lets you get close to the machine
if you really want to.
.PP
If your application requires you to translate
a language into a set of actions or another language,
you are in effect building a compiler,
though probably a small one.
In that case,
you should be using
the
.C yacc
compiler-compiler,
which helps you develop a compiler quickly.