Dec 89 Letters
Volume Number: 5
Issue Number: 12
Column Tag: Letters
By David E. Smith, Editor & Publisher, MacTutor
There IS an OOP Book
James Plamondon
San Mateo, CA
I have just received and read my copy of the September issue of MacTutor, and
was pleased to discover that it was up to its usual high standards. Unfortunately, I was
horrified to see one piece of misinformation presented -- however unintentionally --
as fact, when the truth could be of great use to your readership.
In the article MacOOPs!, by Jean-Denis Muys-Vasovic, on page 102, in the first
paragraph, the author says that “Algorithms + Data Structures = Programs”, by
Niklaus Wirth, was the book that established the case for structured programming in
the s eventies. This is entirely correct.
Unfortunately, the author goes on to say that ‘This technology [OOP] has not yet
got its “Algorithms + Data Structures = Programs”’, meaning that there is not yet
any single book on OOP that explains its concepts, advantages, implementation, and so
on. This statement has not been true for more than a year.
Everyone who has an interest in OOP should run -- not walk -- to the nearest
computer bookstore and buy a copy of Bertrand Meyer’s book, “Object-Oriented
Software Construction”, published in 1988 by Prentice Hall (about $40). It is
without a doubt the most comprehensive, concise, and yet readable book on
Object-Oriented Design and Programming to date -- and I have no doubt that it will
soon be considered to be as important to the success of OOP as Wirth’s book was to
structured programming.
Just as Wirth’s book explained and made the case for block structuring, strong
typing, and functional decomposition, Meyer’s book explains and makes the case for
object-oriented design, based on considerations of correctness, robustness,
extendibility, reusability, and compatibility. Meyer’s book, like Wirth’s, combines
the rigorous precision of computer science with the practical aspects of software
engineering. Wirth used his own, new language, Pascal, to present his ideas, because
the other languages of the time were inadequate. Likewise, Meyer uses his own
language, Eiffel, to present his ideas, because no other existing language so elegantly
addresses both the theoretical and practical issues of object-oriented design.
Pascal went on to become one of the major computer languages of the s eventies
and eighties, due to its close association with structured programming. Its commercial
use was hampered by the fact that it was designed to be a teaching aid, not a language
for software engineering. Eiffel, on the other hand, was designed from the first to be a
practical tool for the development of commercial software. Its future is unlimited.
In every discipline, its practitioners struggle for years on the fringes of great
breakthroughs -- gravity, natural selection, relativity, plate tectonics -- until some
author comes along and puts it all together. That is what Wirth did for structured
programming, and that is what Meyer has done for object-oriented programming. I
was a geologist during the battles that led to the acceptance of plate tectonics, so I’ve
learned to recognize a paradigm shift when I see one.
If knowledge is power, and the future of programming is OOP, then this book will
give you the power to control the future of programming. Take the day off, grab the
company credit card, go to the nearest computer bookstore, buy Meyer’s book, and
don’t listen to another word about OOP until you’re done reading it.
Stop reading this letter! Get moving! Run!
What, you’re still here?
Virus?
George Patail
APO New York
I am new to Mac programming and an avid reader of MacTutor and need your help
regarding Mac viruses and their eradication. My system (a Mac II with an 80MB HD)
had a bad case of the “hangs” lasting for 15-20 minutes on bootup and when launching
applications. I cleared out all the files and reinstalled them after examining each one
with ResEdit. I installed Vaccine on my system and it seems to keep a vigilant eye on
any attempts to add additional CODE resources. It warns me and asks for the usual
permission to let it add or deny access to the CODE resource.
However running LS Pascal 2.0, it asks for permission with the usual message
for every source file I try to add to a project. I launched ResEdit and examined my
System Folder for any telltale signs of Scores or nVIR viruses following Max Rochlin’s
instructions (Mousehole Report, MacTutor June 1988). None were found! However,
the LS Pascal application’s CODE id=0’s el eventh word displays ‘038A’ and the tenth
displays ‘4EED’. According to MaxR this is a sure sign of Scores infection! But I could
not locate any additional CODE segments corresponding to 038A which is two higher
than the next, nor atpl id 128, data id -4001, inits 6, 10, 17, _Desktop and Scores
files in the System Folder. I checked the master LS Pascal 2.0 disks I got from Think
and they too display the same anomalies with 4EED and 038A. Is the LS Pascal
infected, or is it normal code for LS? Do you think it is a new strain? It does not seem
to spread to other applications so far. But I am at a loss as to how to combat this other
than turning Vaccine off, which I don’t want to do knowing if it is safe to run LS Pascal.
I am stationed in Heidelberg Germany and the local Apple reps could not answer
anything beyond the most elementary questions.
[LS Pascal adds code resources to its project file each time you compile, so
naturally Vaccine will intercept the attempt to add a code type resource, like CODEs and
CDEFs. My copy of LS Pascal also contained the same words in CODE 0. I have been
using VIREX by HJC Software and Vaccine before without a hint of virus troubles. A
quick call to Symantec said that the jump table is beyond that point and those are the
words that should be there. So my opinion is that you don’t have a virus. I would
suggest your ‘hangs’ are due to INITs. One of them is likely to be giving you problems.
If you are worried about virus problems, may I suggest that VIREX or another
commercial virus program. It is my understanding that Vaccine is not continually
updated as other commercial programs. -ed]
To Paul Onstad
John Baxter
Port Ludlow, WA
As others have noted, the solution to Paul Onstad’s question about opening a
resource fork in a desired folder is straight forward (e specially for those of us who
have been “around” the Mac since the early days). Simply save the current default
volume with GetVol, then set the default volume to the “ volume” (actually a working
directory in most cases) returned by the SFPutFile call (or earlier SFGetFile call if
replacing a file). Then use the old resource file opening calls without concern. When
completed, set the default volume back.
Furthermore, when rewriting a file as a result of Save, it is not necessary to
erase the old file. It is sufficient to set the EOF of the data fork to zero before writing
the new data. This will leave the resource fork, the creation date, and the position
within the folder as seen by Finder unchanged. Only for Save As should it be
necessary to copy the resource fork. [It’s actually only necessary to set EOF to the new
value, and then only if the old file is larger. if the new EOF value is known before
starting to write the data, this might be preferred.] The SetEOF call takes the file’s
reference number and the desired new EOF value, extending or truncating the file as
needed. Inside Macintosh, Vol. II, pg. 93.
More Paul Onstad
John H. Roth, Jr.
Chicago, IL
Your new cover style almost lost you a reader-I found your August issue on the
magazine rack by accident. For a while I thought you had gone under (perish the
thought).
By now, you’ve probably got a dozen letters about Paul Onstad’s resource fork
problem. The error from OpenRFPerm is because the new resource fork doesn’t have a
resource map. This can be corrected by a call to CreateResFile, after the call to Create
but before the call to OpenRFPerm. The sequence of calls is crucial; odd things can
happen if Create is not called first. This is explained in Tech Note 101.
The basic misconception seems to be that the Create call builds both forks of the
file. Actually, it builds an empty directory entry. For the data fork, Open is then
sufficient, but for the resource fork, initialization data has to be written into it before
the Resource Manager can do its thing.
THINK C 4.0 Bugs
William Modesitt
Kaneohe, HI
There are two known bugs in the new THINK C 4.0. Both involve scanf. When
using the standard library ANSI, scanf will not work correctly for float values less
than .1. When using the library ANSI_881, scanf in general does not work correctly
for float values. I was told by THINK C technical staff that these were the only known
bugs, and they were both rather nasty. The fixes are shown below.
Buggy version:
case -2:
if (c >= ‘0’ && c <= ‘9’) {
F.valid = TRUE;
if (c != ‘0’ || D.sig[0]) {
if (D.sig[0] < sizeof D.sig - 1)
D.sig[++D.sig[0]] = c;
if (F.dot)
--D.exp;
}
}
Correct version:
/* 1 */
if (c >= ‘0’ && c <= ‘9’) {
F.valid = TRUE;
if (c != ‘0’ || D.sig[0]) {
if (D.sig[0] < sizeof D.sig - 1)
D.sig[++D.sig[0]] = c;
} /* paren was moved here from two lines below */
if (F.dot)
--D.exp;
}
Buggy version (881):
register struct decrec *d;
void *p;
{
short excp class, c = ‘0’;
#ifdef _MC68881_
register short *q = p;
p = q + 1;
#endif
Correct version (881):
register struct decrec *d;
void *p;
{
short excp class, c = ‘0’;
#ifdef _MC68881_
register short *q = p;
if (code == FFEXT) /* add this line */
p = q + 1;
#endif
Buggy version (881):
#ifdef _MC68881_
q[0] = q[1];
q[1] = 0;
#endif
return;
Correct version (881):
/* 2 */
#ifdef _MC68881_
if (code == FFEXT) { /* add this line */
q[0] = q[1];
q[1] = 0;
} /* add this line */
#endif
return;
Once these changes have been made, you should “make” all four ANSI libraries by
choosing the make option, check scanf.c, then “make” it. Any projects that you have
will automatically load the new ANSI library for you next time you update.
And now for something completely different. I have tried to find a way to
determine the current volume reference number or working directory reference
number for my current application during run time so I can write to it’s data fork.
Along the same line, I think, I would like to do the following: Implement SFGetFile,
exit, find out where I am, move to the parent directory, and implement SFGetFile again
in the new (parent) directory. What I need is a function called “WhereAmI”. Any
help would be greatly appreciated.
Linear Equation Error
Wilfred Long, Jr.
Montgomery Village, MD
In C. H. Friend’s Linear Equations True Basic examples from January’s letters
column, he couldn’t produced the output shown in the Multiple Linear Regression”
segment by running the source as listed. All the errors are on page 109 in the last
third of the program following the REM statement depicting how H-P Basic allows
solution of the linear system.
In the MAT PRINT USING formatter for “Coefficients” there should be three
leading dashes instead of two and the comma should be a period.
The LET statement in the vector dot-product loop following “residuals:” is
incomplete-no product is indicated. It should be written either:
LET Y2(I) = Y(I,1) * Y(I,1)
or
LET Y2(I) = Y(I,1)^2
In the second from the last PRINT USING statement, the variable SSR should be
SSA in both instances. There is no variable SSR, but SSA produces the value 399.454
shown in the last results-output line attributed to SSR.
Finally, in the last PRINT USING formatter for variable ES2, the comma again
should be a period.
There are a couple of incorrect values in the results-output, but they’ll be
obvious when the corrected program runs.
Fulfilling A Dream
Natalia Kolodziejska
Wroclaw, Poland
I am an 11-year old Polish schoolgirl dreaming of an Apple Macintosh.
Poland is full of personal computers. There are many Atari, Commodore, IBM
and Sinclair/Timex computers.
There also are other computers in use. However, the Apple Macintosh is absent
in Poland. Anyway, I was lucky enough to see and use this computer some times. I am
sure it would be my best friend.
However, I have a very little chance to come into its possession by myself. I
simply lack the proper means. So, perhaps somewhere in the world there are people
or institutions that have one superfluous (used) Macintosh which could be donated to
me. (I would also welcome a printer and some software.) Please forward this letter to
your readers and friends. Everyone can contact me, even if he or she can help me to
only some extent.
Using the Macintosh I could better prepare myself for life in the next century.
My best greetings,
Natalia Kolodziejska
ul. Drzewieckiego 16/4
PL 54-129 Wroclaw
Poland