Get 1989/westley to compile with clang

Some important notes here should be made. First of all with clang the
recommended commands in the README.md (which was also formatted and typo
fixed) will seemingly generate the correct files but still only
westley.c and ver0.c will compile with clang. With gcc they do compile
and seem to work as expected.

However I found a bug in the original code too where ver2 will sometimes
segfault. This also goes for the fix (again when compiled with gcc).

I did get ver[123].c to also compile but this seemed to cause invalid
output in some of the commands. That being said it might (though I don't
think so at this time) have been not knowing exactly what should be
displayed. Maybe I will look at it in the future but at least it now
works with clang even if not all of the generated files will compile
that way. Still macOS users can now take a look at the different output
of the program which seems good to me (it does not even have to be
source code that is output).

These things have been noted in the README.md as well.
This commit is contained in:
Cody Boone Ferguson committed 2023-03-17 06:55:06 -07:00
1 parent 45fd90259b
commit 892b94c5a6
4 files changed
+117 -79

No files matched your search

+8
View File
@@ -1 +1,9 @@
ver0
ver0.c
ver1
ver1.c
ver2
ver2.c
ver3
ver3.c
westley
+1 -1
View File
@@ -94,7 +94,7 @@ ARCH=
# Defines that are needed to compile
#
#CDEFINE=
CDEFINE= -Dtrgpune=putchar
CDEFINE= -Dtrgpune=putchar -Dpune=char -Dvag=int
# Include files that are needed to compile
#
+105 -76
View File
@@ -1,4 +1,4 @@
Most algorithms in one program:
# Most algorithms in one program:
Merlyn LeRoy (Brian Westley)
Starfire Consulting
@@ -7,117 +7,146 @@ Most algorithms in one program:
55104
USA
Judges notes:
## To build:
There is a secret way to get ONE of the versions to print out
"Hello, world!\n". Can you find how to do it?
make all
Try the following commands:
westley < westley.c > ver0.c
westley 1 < westley.c > ver1.c
westley 1 2 < westley.c > ver2.c
westley 1 2 3 < westley.c > ver3.c
Try compiling and running the 4 resulting programs.
NOTE: [Cody Boone Ferguson](/winners.html#Cody_Boone_Ferguson) was able to get
this to partially work with clang; clang requires that the second, third and
fourth args to main() be `char **`. Only `westley.c` and `ver0.c` (see below) will
compile with clang. It's possible to get ver1.c to compile too but this causes
problems with generating the output of the remaining files so we have opted, at
least for now, to not do that, as it does not have to output just source code.
He notes that the original source code sometimes segfaults with `ver2` which
also happens with the fix but it'll only compile with a compiler that
does not require the second, third and fourth args of main() to be a `char **`.
Thank you Cody for your help!
Selected notes from the author:
This is a filter. If it is run with no arguments, it copies
stdin to stdout. With one argument, it ROT13's stdin to
stdout. With two arguments, it reverses stdin to stdout. With
three arguments, it does both. It requires the ASCII character
set, with all characters being in 0..255, and EOF must be -1.
Also requires two's complement (I think).
## Try:
The source code will run if ROT13'ed and/or reversed, using a
different algorithm for each version (hereafter referred to as
ver0 (original), ver1 (ROT13), ver2 (reversed), and ver3
(ROT13 and reversed)).
When compiling these versions, one needs to define 'trgpune'
in the compile line. Example:
./westley < westley.c > ver0.c
./westley 1 < westley.c > ver1.c
./westley 1 2 < westley.c > ver2.c
./westley 1 2 3 < westley.c > ver3.c
Try compiling and running the 4 resulting programs. To compile try:
make ver0
make ver1
make ver2
make ver3
## Judges notes:
There is a secret way to get ONE of the versions to print out
"Hello, world!\n". Can you find how to do it?
## Selected notes from the author:
This is a filter. If it is run with no arguments, it copies
stdin to stdout. With one argument, it ROT13's stdin to
stdout. With two arguments, it reverses stdin to stdout. With
three arguments, it does both. It requires the ASCII character
set, with all characters being in 0..255, and EOF must be -1.
Also requires two's complement (I think).
The source code will run if ROT13'ed and/or reversed, using a
different algorithm for each version (hereafter referred to as
ver0 (original), ver1 (ROT13), ver2 (reversed), and ver3
(ROT13 and reversed)).
When compiling these versions, one needs to define 'trgpune'
in the compile line (which the Makefile now has). Example:
cc -Dtrgpune=putchar ver3.c -o ver3
"trgpune" is the ROT13 of getchar(), so getchar() and putchar()
are exchanged in the ROT13 counterpart. It is easy to see that
this is unavoidable. I must have a #define for a library
function; otherwise I would have an unidentifed extern for the
ROT13 version. If I then define this function, it won't link
in the library version for the ORIGINAL code, since my
definition will supercede the library function. Hence, the
compiler option gives me putchar(), and allows me to use
getchar(). I pass a dummy argument to getchar() to eliminate
"variable number of args" from lint (unless it checks against
the library). Otherwise, all versions lint reasonably (main
returns random value & constant in conditional context [when I
check for ROT13 version] is all it complains about).
"trgpune" is the ROT13 of getchar(), so getchar() and putchar()
are exchanged in the ROT13 counterpart. It is easy to see that
this is unavoidable. I must have a #define for a library
function; otherwise I would have an unidentified extern for the
ROT13 version. If I then define this function, it won't link
in the library version for the ORIGINAL code, since my
definition will supersede the library function. Hence, the
compiler option gives me putchar(), and allows me to use
getchar(). I pass a dummy argument to getchar() to eliminate
"variable number of args" from lint (unless it checks against
the library). Otherwise, all versions lint reasonably (main
returns random value & constant in conditional context [when I
check for ROT13 version] is all it complains about).
ver0 and ver1 use a range check and a calculation to do ROT13,
while ver2 and ver3 use table lookup. All versions contain
main() and it's ROT13 fn, znva(). ver0/ver1 [ver2/ver3] are
(of course) syntactically identical, since the syntax is in the
non-alphabetic characters. However, since one program starts
at main() while it's ROT13 counterpart starts at znva(), znva()
calls main (znva() is also used for output).
ver0 and ver1 use a range check and a calculation to do ROT13,
while ver2 and ver3 use table lookup. All versions contain
main() and it's ROT13 fn, znva(). ver0/ver1 [ver2/ver3] are
(of course) syntactically identical, since the syntax is in the
non-alphabetic characters. However, since one program starts
at main() while it's ROT13 counterpart starts at znva(), znva()
calls main (znva() is also used for output).
All versions use recursion to work. If the program is NOT
reversing it's output, it prints out the (possibly ROT13'd)
character before recursing, otherwise it prints it out
afterward (or doesn't recurse at all when EOF is reached).
Since most of this code is identical, it is put into znva() and
called with a first parameter of 0 as a flag (as "main()", it's
first argument (argc) must be at least one).
All versions use recursion to work. If the program is NOT
reversing it's output, it prints out the (possibly ROT13'd)
character before recursing, otherwise it prints it out
afterward (or doesn't recurse at all when EOF is reached).
Since most of this code is identical, it is put into znva() and
called with a first parameter of 0 as a flag (as "main()", it's
first argument (argc) must be at least one).
I can't use any flow control. If I used if(), I would have a
function vs() to define in the ROT13 version. But a function
called vs() turns into a function called if() in the original,
so it can't be done. Therefore, I do:
I can't use any flow control. If I used if(), I would have a
function vs() to define in the ROT13 version. But a function
called vs() turns into a function called if() in the original,
so it can't be done. Therefore, I do:
expr1 && expr2 && (expr3=etc);
which is the same as:
which is the same as:
if (expr1 && expr2) expr3=etc;
A/UX on the Macintosh doesn't get this right; it evaluates ALL
expressions if they aren't in an assignment or conditional
statement. This might warrant a warning, since other compilers
may do this. I found MANY compilers botched:
A/UX on the Macintosh doesn't get this right; it evaluates ALL
expressions if they aren't in an assignment or conditional
statement. This might warrant a warning, since other compilers
may do this. I found MANY compilers botched:
expr1 && (expr2,expr3);
expr2 was OFTEN evaluated even if expr1 was false. I removed
such statements to make it more portable.
expr2 was OFTEN evaluated even if expr1 was false. I removed
such statements to make it more portable.
The variable names are worth noting:
The variable names are worth noting:
'irk' and 'vex' are ROT13 pairs and are synonyms.
'Near' and 'Arne' are ROT13 pairs and are anagrams.
'NOON' and 'ABBA' are ROT13 pairs and are palindromes.
'tang' and 'gnat' are both ROT13 and palindrome pairs!
Normally (!), a reversible C program is done thus:
Normally (!), a reversible C program is done thus:
/**/ forward code /*/ edoc drawkcab /**/
If your compiler nests comments, it will get this wrong.
However, I have made some bits of the code palindromic,
(or different, but reversible) so it is more like:
If your compiler nests comments, it will get this wrong.
However, I have made some bits of the code palindromic,
(or different, but reversible) so it is more like:
/**/forward/*//**/ palindromic /**//*/drawkcab/**/
The code can therefore be interlaced. There are eight
such palindromic bits. You can find them within the
/*//**/ /**//*/ pairs.
The code can therefore be interlaced. There are eight
such palindromic bits. You can find them within the
`/*//**/ /**//*/` pairs.
The body of the code of ver0 and ver1 is a large lumpy 'K' (for
Kernighan); the code of ver2 and ver3 is a flat-topped and
lumpier 'R' (for Ritchie). Judicious use of spaces and tabs
helped here. It barely fits on an 80x24 screen. Squint. Note
that the code must start with a blank line, or the reversed version
will lack a terminating newline.
The body of the code of ver0 and ver1 is a large lumpy 'K' (for
Kernighan); the code of ver2 and ver3 is a flat-topped and
lumpier 'R' (for Ritchie). Judicious use of spaces and tabs
helped here. It barely fits on an 80x24 screen. Squint. Note
that the code must start with a blank line, or the reversed version
will lack a terminating newline.
Copyright (c) 1989, Landon Curt Noll & Larry Bassel.
All Rights Reserved. Permission for personal, educational or non-profit use is
+3 -2
View File
@@ -1,7 +1,8 @@
/**//*/};)/**/main(/*//**/tang ,gnat/**//*/,ABBA~,0-0(avnz;)0-0,tang,raeN
,ABBA(niam&&)))2-]--tang-[kri - =raeN(&&0<)/*clerk*/,noon,raeN){(!tang&&
noon!=-1&&(gnat&2)&&((raeN&&( getchar(noon+0)))||(1-raeN&&(trgpune(noon
,ABBA(niam&&)))2-]--tang-[kri - =raeN(&&0<)/*clerk*/,noon,raeN)char
**gnat, **noon, **raeN;{int gn = gnat, ra = raeN, no = noon; (!tang&&
no!=-1&&(gn&2)&&((raeN&&( getchar(no+0)))||(1-ra&&(trgpune(no
)))))||tang&&znva(/*//**/tang ,tang,tang/**|**//*/((||)))0(enupgrt=raeN
(&&tang!(||)))0(rahcteg=raeN( &&1==tang((&&1-^)gnat=raeN(;;;)tang,gnat
,ABBA,0(avnz;)gnat:46+]552&)191+gnat([kri?0>]652%)191+gnat([kri=gnat