This prac is due for submission by lunch time on your next practical day, correctly packaged in a transparent folder as usual (unpackaged and late practical submissions will not be accepted - you have been warned). Pracs should please be deposited in the hand-in box outside the lab. Only one set of listings is needed for each group, but please enclose as many copies of the cover sheet as are needed, one for each member of the group. These will be returned to you in due course.
In this practical you are to
The exercises for this week are not really difficult, although they may take longer than they deserve simply because you may be unfamiliar with the systems.
Copies of this handout, the cover sheet, the Parva language report, and descriptions of the library routines for input, output, string handling and set handling in Java and C# are available on the course web site at http://www.cs.ru.ac.za/CSc301/Translators/trans.htm.
When you have completed this practical you should understand
This week your group is required to hand in, besides the individual cover sheets for each member:
Keep the cover sheet and your solutions until the end of the semester. Check carefully that your mark has been entered into the Departmental Records.
You are referred to the rules for practical submission which are clearly stated in our Departmental Handbook. However, for this course pracs must be posted in the "hand-in" box outside the laboratory before the next practical session and not given to demonstrators during the session.
A rule not stated there, but which should be obvious, is that you are not allowed to hand in another student's or group's work as your own. Attempts to do this will result in (at best) a mark of zero and (at worst) severe disciplinary action and the loss of your DP. You are allowed - even encouraged - to work and study with other students, but if you do this you are asked to acknowledge that you have done so on all cover sheets and with suuitable comments typed into all listings. You are expected to be familiar with the University Policy on Plagiarism, which you can consult at:
http://www.scifac.ru.ac.za/plag.htm
In this practical course you will be using a lot of simple utilities, and usually work at the "command line" level rather than in a GUI environment. Note in particular:
LPRINT Fibo.pav Fibo.java
The listings come out in a small font which enables long lines to be read easily and with narrow line spacing (so that you get more listing for your money). Please use this utility or the standard "2 up" UltraEdit configuration to produce all listings submitted on this course, as it makes my job of reading the submissions much easier. Program listings in "proportional font" are awkward to read.
For this prac it is recommended that you simply work in the Hamilton lab, rather than begging, borrowing or stealing copies of a whole host of software for home use. In future pracs you will mostly use Java only, and the prac kits will contain nearly all the extras you need.
We shall make use of zipped prac kits throughout the course; you will typically find sources for each week's prac in a file pracNN.zip on the server. Copy prac19.zip and xtacy.zip needed for this week, either directly from the server on I:\CSC301\TRANS (or by using the WWW link on the course page), and extract the sources when you need them, into your own directory/folder, by using UNZIP.
copy i:\csc301\trans\prac19.zip
unzip prac19.zip
Use UNZIP or WINZIP and not PKUNZIP, as the file contains files with long file names which PKUNZIP cannot handle.
In the past there has been a problem with running applications generated by the C# compiler if these are stored on the network drives. This may not yet have been completely resolved, so for those parts of the practical that involve the use of C#, work from the local D: drive instead. After opening a command window, log onto the D: drive, create a working directory and unpack a copy of the prac kit there:
d:
md d:\G01T1111
cd d:\G01T1111
unzip I:\csc301\trans\prac19.zip
In the prac kit you will find various versions of a famous program for finding a list of prime numbers using the method known as the Sieve of Eratosthenes. You will also find various versions of a program for computing Fibonacci numbers recursively, some "empty" programs, and some other bits and pieces, including a few batch files to make some of the following tasks easier.
You may not be a Pascal expert, but in the kit you will find three Pascal programs, including SIEVE.PAS that determines prime numbers using a Boolean array to form a "sieve". Study and compile these programs - you can do this from the command line quite easily by issuing commands like
FPC SIEVE.PAS
FPC FIBO.PAS
FPC EMPTY.PAS
to use the 32-bit Windows version of the Free Pascal compiler. Make a note of the size of the executable (use the command DIR SIEVE.EXE and DIR FIBO.EXE and DIR EMPTY.EXE).
You may be able to produce a slightly faster version of the executable program for the Sieve example by suppressing the index range checks that Pascal compilers normally include for code that accesses arrays:
FPO SIEVE.PAS
How do the sizes of the executables compare? Why do you suppose the "empty" program produces the amount of code that it does?
Here is something more demanding: By experimenting with the CONST declaration, find out how large a sieve the program can handle. What is the significance of this limit? Hint: you should find that funny things happen when the sieve gets too large, though it may not immediately be apparent.
You may not be a Modula-2 expert either, but examine, and then compile and run the equivalent Modula-2 code supplied in the files SIEVE.MOD, EMPTY.MOD and FIBO.MOD. You can do this quickly using commands like
M2C FIBO (note that the .MOD extension is not quoted here) or
M2O SIEVE (for the version that suppresses subscript checks)
Make a note of the size of the executables produced. How do they compare with the Pascal executables? Approximately how big a sieve can the compiler handle? Why do you suppose there is a difference, when the source programs are all so similar?
The kit also includes C and C++ versions of these programs. Compile these and experiment with them in the same way:
BCC SIEVE.C (using the Borland compiler in C mode)
BCC SIEVE.CPP (using the Borland compiler in C++ mode)
CL FIBO.C (using the WatCom compiler in C mode)
CL FIBO.CPP (using the WatCom compiler in C++ mode)
Once again, make a note of the size of the executables, and in particular, compare them with the earlier versions. Can you think of any reason why the differences are as you find them?
There are two Java compilers available for your use. The JDK one is called javac and there is also the (much faster) one called jikes (Jikes will only handle Java 1.4 level source, but that covers most things). Both of these are conveniently invoked from within UltraEdit. You can also compile a Java program directly from the command line with commands like
javac Sieve.java (using the (slow) JDK compiler)
jikes Sieve.java (using the (fast) Jikes compiler)
You can compile the C# versions of these programs from the command line, for example:
csharp Sieve.cs
(You may have to do this on the local D: drive) Make a note of the size of the ".NET assemblies" produced (SIEVE.EXE, EMPTY.EXE and FIBO.EXE). How do these compare with the other executables?
On the course web page you will find a description of Parva, a toy language very similar to C, and a language for variations on which we shall develop a compiler and interpreter later in the course. The main difference between Parva and C/Java/C# is that Parva is stripped down to bare essentials.
Learn the Parva system by studying the language description where necessary, and trying the system out on the supplied code (SIEVE.PAV and FIBO.PAV). There are various ways to compile Parva programs. The easiest is to use a command line command:
parva Sieve.pav simple error messages
parva -o Sieve.pav slightly optimized code
parva -l Sieve.pav error messages merged into listing.txt
You will have to do this on the local D: drive.
More conveniently, we have set up UltraEdit to allow for an option to compile Parva programs. If you want to add this to your home systems, use the Advanced->Tool Configuration pull down, then set the following fields
Command Line Parva %n%e
Working Directory %p
Menu Item Name Parva
Save all files first Selected
Output to List Box Selected
Capture Output Selected
and then click Insert. After this you can choose the Parva option on the Advanced menu to compile (and, when successful, run) the program in the "current window". The demonstration programs Sieve.pav and Fibo.pav in the kit have a few fairly obvious errors. Learn the syntax and semantics of Parva by correcting the errors until the programs run correctly. Once again, experiment to see how large a sieve you can set up.
It may help amplify the material we are discussing in lectures if you put some simple Modula-2 programs through a high-level translator we have available, and then look at, and compile, the C code to see the sort of thing that happens when one performs automatic translation of a program from one high-level language to another.
We have a demonstration copy of a system (Russian in origin), that translates Modula-2 or Oberon-2 source code into C. The system is called Extacy (a poor pun on "X to C", it seems). Whether or not the C one obtains is usable depends, obviously, on having C translations of all of one's Modula-2 libraries as well. In principle all one has to do is convert these libraries using the same system. Some very simple libraries came with the demonstration kit, and we have produced one or two more, but we would have to pay many Roubles and do an awful lot of work to get the system fully operational.
XC =m SOURCE.MOD
will produce all the .H and .C files needed for a "make" of the parent program SOURCE.MOD
Convert the sample programs in this kit (SIEVE.MOD and FIBO.MOD) and the various support modules to C, and then use a C++ compiler to compile and run the resulting code. Most simply, run the C compiler directly from the command line:
BCC SIEVE.C EASYIO.C X2C.C
BCC FIBO.C EASYIO.C X2C.C
or
CL SIEVE.C EASYIO.C X2C.C
CL FIBO.C EASYIO.C X2C.C
Take note of, and comment on, such things as the kind of C code that is generated (is it readable; is it anything like you might have written yourself?), and of the relative ease or difficulty of using such a system. You might also like to comment on the size of the object code produced.
Different compilers - even for very similar programs - can produce code of very different quality. In particular "interpretive" systems (of which the Parva implementation is one example) produce programs that run far more slowly than do "machine" or "native" code systems. Carry out some tests to see these effects for yourselves, and how severe they are, by comparing the execution times of some of the programs.
Hint: the machines in the Hamilton Labs are very fast, so you should try something like this: modify the sieve programs to comment out nearly all the output statements (since you are not interested in seeing a zillion lists of prime numbers, or measuring the speed of I/O operations), and then run the programs and time them with a stop watch. Similarly, choose an upper limit for the list of Fibonacci numbers (and a suitable number of iterations and sizes for the sieve) that will produce measurable times.
Although Java is often touted as being an interpreted language, in fact the latest versions of the Java "interpreter" - the program executed when you give the java command - actually indulge in "just in time" compiling (see textbook page 32) and "JIT" the code to native machine code as and when it is convenient - which results in spectacularly improved performance. It is possible to frustrate this by issuing the java command with a directive -Xint:
javac Sieve.java
java -Xint Sieve
to run the program in interpretive mode. Try this out as part of your experiment.
Summarise your findings on page 2 of the cover sheet, and go on to explain briefly how you come to the figures that you quote. For example, is Java better/worse than C# (the source code in each case is almost identical)? Do 16-bit compilers fare better or worse than 32-bit compilers?
In lectures you were told of the existence of decompilers - programs that can take low-level code and attempt to reconstruct higher level code. There are a few of these available for experiment.
jad a decompiler that tries to construct Java source from Java class files
javap a decompiler that creates pseudo assembler source from a Java class file
gnoloo a decompiler that creates JVM assembler source from a class file
oolong an assembler that creates Java class files from JVM assembler source
ildasm a decompiler that creates CIL assembler source from a .NET assembly
ilasm an assembler that creates a .NET assembly from CIL assembler source
peverify a tool for verifying .NET assemblies
Try out the following experiments or others like them:
(a) After compiling Sieve.java to create Sieve.class, decompile this:
jad Sieve.class
and examine the output, which will appear in Sieve.jad
(b) Disassemble Sieve.class
javap -c Sieve >Sieve.jvm
and examine the output, which will appear in Sieve.jvm
(c) Disassemble Sieve.class
gnoloo Sieve.class
and examine the output, which will appear in Sieve.j
(d) Reassemble Sieve.j
oolong Sieve.j
and try to execute the resulting class file
java Sieve
(e) Be malicious! Corrupt Sieve.j - simply delete a few line with opcodes on them. Try to reassemble the file (as above) and to re-run it. What happens?
(f) Compile Sieve.cs and then disassemble it
csharp Sieve.cs
ildasm /OUT=Sieve.cil Sieve.exe
and examine the output, which will appear in Sieve.cil
(g) Reassemble Sieve.cil
ilasm Sieve.cil
and try to execute the resulting class file
Sieve
(h) Be malicious! Corrupt Sieve.cil - simply delete a few line with opcodes on them. Try to reassemble the file (as above) and to re-run it. What happens?
(i) Experiment with the .NET verifier after (h)
peverify Sieve.exe
Nothing you have done so far should have extended your programming talents very much. To get the old brain cells working a little harder, turn your minds to the following. No - the idea is not to farm the problems out one to each member of the group; argue about them together!
Please read the next section carefully before you begin:
You will need to become acquainted with various library classes to solve these problems. A simple sample program using some of the library routines can be found in the kit as the program SampleIO.java, which is listed below. Descriptions of these and other library routines can be found on the course website.
Please don't just rush into them, and please don't just "do your own thing". The exercise is supposed to familiarize you with libraries you will use again later in the course. In particular the I/O must be done using the IO, InFile and OutFile classes and not all sorts of other classes and I/O and parsing methods that you might know of or read up about. Study the SampleIO program carefully and familiarize yourself with routines like readChar, readString and readWord.
It is important that you learn to use the IO libraries InFile, OutFile and IO. These will be used repeatedly in this course. Please do not use other methods for doing I/O, or spend time writing lots of exception handling code.
Pat Terry's problems are sometimes reputed to be hard. They only get very hard if you don't think very carefully about what you are trying to do, and they get much easier if you think hard and spend time discussing the solutions with the tutors or even the Tyrant himself. His experience of watching the current generation of students suggests that some of you get beguiled by glitzy environments and think that programs just "happen" if you can guess what to click on next. Don't just go in and hack. It really does not save you any time, it just wastes it. Each of these problems can be solved elegantly in a small number of lines of code if you think them through carefully before you start to use the editor, and I shall be looking for elegant solutions.
Remember a crucial theme of this course - "Keep it as simple as you can, but no simpler".
The first problem calls for the use of simple IO (use the InFile and OutFile classes, and in particular the readWord method) and simple tokenizing (use the StringTokenizer class).
On the Internet, e-mail addresses are typically of the form user@machine.site.subdomain.country -for example p.terry@cs.ru.ac.za. In the UK, the older Janet network "drove on the wrong side of the road" - the addresses were quoted as username@uk.ac.site.machine. Write a program that will process a list of Janet-style addresses to produce a list of Internet-style addresses, and/or vice versa.
This exercise has various elements of actual compilation - you have to be able to decode a string into simpler tokens and then analyze this stream of tokens before creating tokens that are copied to an appropriate output stream. Assume, for simplicity, that the list of addresses will be correct, and that the components of the address and user name can contain only letters, digits and hyphens. Speciment data is in the files ADDRESSES.
Consider a sequence originally containing the N ordinal numbers
1 2 3 4 5 6 7 8 9 10 11 12 13 14 ... N
Removing every second number produces the new sequence
1 3 5 7 9 11 13 15 17 19 21 23 25 27 ...
and now removing every third number from that sequence generates another new sequence
1 3 7 9 13 15 19 21 25 27 ...
and removing every fourth number from this sequence produces another new sequence
1 3 7 13 15 19 25 27 ...
This process continues indefinitely, or more exactly, as long as numbers are still there to be removed. For example if N = 12 the process would terminate once we had generated the sequence 1 3 7.
The numbers which survive the culling process indefinitely are deemed to be "lucky".
Write a Parva program to determine the lucky numbers less than some number provided as input data.
Hint: Write a function that will generate a list of such numbers in an array parameter when supplied with an upper limit N, and use this in a program that displays such a list.
A photographer is about to photograph a class of students (yes, we know, Barns took ages to get it right) and wants them neatly arranged by height. The plan is to have the taller people in the back rows, and for each row to slope down from the centre towards each edge. Given a class of students and their heights, Barns needs a plan of where each should stand. The program should choose the number of rows intelligently, depending on the size of the class.
You can find an executable version of a program that does this (PHOTO.EXE), and a data file from which you could select the first N students (STUDENTS) in the prac kit to try out your ideas.
Remember a crucial theme of this course - "Keep it as simple as you can, but no simpler".
A recurrent problem that students have is to check timetables for clashes. Wouldn't it be nice if we had programs that could do this for us?
Well, of course we have, if we are prepared to write them. In the prac kit you will find a data file that describes the Rhodes Timetable, and a program CLASH.EXE (originally written in C#; you will have to run it from the D: drive) that reads this file, and then allows you to type in pairs of subjects, for each of which it will report whether there are any clashes, and when these occur. A session with the program might go something like this
D:>CLASH Subject 1 (or STOP) ? csc 1 Subject 2 (or STOP) ? gog 2 CSC 1 Mon 1 Tue 2 Wed 3 Thu 4 Fri 5 GOG 2 Mon 3 Tue 4 Wed 5 Wed 7 Wed 8 Wed 9 Wed 10 Thu 1 Fri 2 0 clashes Subject 1 (or STOP) ? csc 1 Subject 2 (or STOP) ? gog 1 CSC 1 Mon 1 Tue 2 Wed 3 Thu 4 Fri 5 GOG 1 Tue 2 Wed 3 Thu 4 Fri 5 4 clashes Tue 2 Wed 3 Thu 4 Fri 5 Subject 1 (or STOP) ? stop D:>
Your biggest task this week is to write a Java program that achieves the same effect.
Warning - although this exercise is actually quite easy, I suspect that this may be a non-trivial problem for many of you. So make sure you understand what is required before you rush in to writing code. Discuss the approach you will take with the demonstrator team, and with each other.
The program involves various aspects of reading and parsing data - that is, recognizing components of data buried in a large input file and extracting them for further processing. Parsing is a fundamental component of programming language translation as well.
This particular problem is neatly solved if you make use of sets - each subject has a set of fixed time table slots (and a possible set of alternative time-table slots, but we can ignore those; you can also ignore any periods out of the range 1 ... 10 in any day). So the program can sensibly make use of an array of suitable records or structures, one field of which is a set of integers.
Part of the data file is listed out below. You will need to study it before you begin. Notice that there is a lot of information on each line that is effectively ignorable for this exercise.
11 July 2008 ; date of last modification ; ; Data file used by the CLASH program 2005 (all semesters too) ; and by WHEN, VENUE, VENALL and WHERE ; ; Format is as follows - spacing important before the "periods" start ; ; Line starting in column 1 is just commentary. Blank lines not allowed!! ; ; Sem = blank (both) or 1/F (1st) or 2/S (second) ; 1 and 2 for courses only offered in one semester ; F and S for components of a year course ; A and B for components of a year course for venues only ; ; * or + are subjects ignored by DUMPVEN and DUMPTIM ; col 2 non blank if you want to highlight changes ; ; Mnemonic of 7 chars is the one that is typed into Pat's programs ; Mnemonic of 7 chars is the one that is useful for venue loading/display ; Protea course code ; Mnemonic of 7 chars is the one that is used by Protea programs ; Followed by guesstimate of size and 5-letter venue code ; ; counts 3 or >4 are used in the venue programs (mixed venue subjects need a count of 2) ; counts 3 and 4 mean the subject is ignored when testing for clashes in Pat's helper options ; counts 2 or >4 are used in the timetable print program (only for blank, 1 or 2 type subjects) ; ; List of timetabled periods followed, terminated with . ; This is a little cryptic, but not too bad once you get used to it ; L = alternative lecture ; T = alternative tut ; A = alternative prac ; P = fixed prac ; no letter - fixed period ; ; 11 = day 1 period 1 ; 35 = day 3 period 5 ; 59 = day 5 period 9 ; 1A = day period 10 ; ; Mnemonic Booked Code Protea Size Venue Full Name Cryptic stuff with . at end ;2345678901234567890123456789012345678901234567890123456789012345678901234567890 ; vvvvvvv vvvvvvv vvvvvvv vvvvvvv vvv vvvvv vvvvvvvvvvvvvvvvvvvvvvvvvvvv vvvvvvvvvvvvvvvvv *vACC 1 ACC 1 4101100 ACC 1 2 ED/ZO Accounting 1 L23 L24 L33 L35 L42 L46 A17 A18 A27 A28 A47 A48 A57 A58 . F ACC 101 ACC 101 4101101 ACC 101 3 ED/ZO Accounting 101 L23 L24 L33 L35 L42 L46 A17 A18 A27 A28 A47 A48 A57 A58 . * CSC 1 CSC 1 5101100 CSC 1 1 ZooMA Computer Science 1 11 22 33 44 55 A17 A18 A19 A1A A27 A28 A29 A2A A37 A38 A39 A3A A57 A58 A59 A5A . *vCSC 2 CSC 2 5101200 CSC 2 5 Glg11 Computer Science 2 13 24 35 41 52 P37 P38 P39 P3A . * CSC 3 CSC 3 5101300 CSC 3 5 Glg11 Computer Science 3 12 23 34 45 51 P47 P48 P49 P4A . DRA 1 DRA 1 2301100 DRA 1 20 Gog11 Drama 1 22 33 55 A19 A1A A29 A2A A39 A3A A49 A4A . A EAR 101 EAR 101 26011G1 EAR 101 4 Eco B Earth Science 101 (Sem 1) 22 33 44 55 A17 A18 A19 A1A A57 A58 A59 A5A P27 P28 P29 P2A . ; ; These next are not really "subjects" but are useful in marking special ; periods when fiddling + MON 1 MON 1 1 11 . + TUE 1 TUE 1 1 21 . + WED 1 WED 1 1 31 . + MON 10 MON 10 1 1A .
You may also like to experiment with the program WHEN.EXE in the prac kit, which creates a graphical output of a timetable for selected subjects and explores whether apparent clashes can be resolved. It does this by using a "backtracking" approach similar to ones you might have seen in other courses. The file KEY summarizes the mnemonic codes used for choosing subjects in the WHEN and CLASH programs.
import java.util.*
; import Library.*;
class SampleIO {
public static void main(String[] args) {
// check that arguments have been supplied
if (args.length != 2) {
IO.writeLine("missing args");
System.exit(1);
}
// attempt to open data file
InFile data = new InFile(args[0]);
if (data.openError()) {
IO.writeLine("cannot open " + args[0]);
System.exit(1);
}
// attempt to open results file
OutFile results = new OutFile(args[1]);
if (results.openError()) {
IO.writeLine("cannot open " + args[1]);
System.exit(1);
}
// various initializations
int total = 0;
SymSet mySet = new SymSet();
// next code is clumsy, but works!
SymSet smallSet = new SymSet(new int[] {1, 2, 3, 4, 5});
String smallSetStr = smallSet.toString();
// read and process data file
int item = data.readInt();
while (!data.noMoreData()) {
total = total + item;
if (item > 0) mySet.incl(item);
item = data.readInt();
}
// write various results to output file
results.write("total = ");
results.writeLine(total, 5);
results.writeLine("unique positive numbers " + mySet.toString());
results.writeLine("union with " + smallSetStr
+ " = " + mySet.union(smallSet).toString());
results.writeLine("intersection with " + smallSetStr
+ " = " + mySet.intersection(smallSet).toString());
} }