mpdevine@watdaisy.UUCP (Michel P. Devine) (06/13/86)
During my CS340 lectures (a second course in Data Structures), I alluded to
some of the unusual features of some programming languages, in particular Pascal
and Fortran. On the spur of the moment (as an (spectacularly unsuccessful)
attempt at stimulating class participation), I suggested an exercise: write a
program which prints the identity of the compiler used to compile it. In the
following script session, both liar.f and liar.p are links to the data file liar
(which is the program in question).
Script started on Mon Jun 9 11:52:23 1986
%
% ls -l liar*
-rw-r--r-- 1 mpdevine 440 Jun 9 11:50 liar
lrwxr-xr-x 1 mpdevine 4 Jun 9 11:45 liar.f@ -> liar
lrwxr-xr-x 1 mpdevine 4 Jun 9 11:45 liar.p@ -> liar
%
% pc liar.p
% a.out
Compiled by a Pascal compiler
%
% f77 liar.f
liar.f:
MAIN liar:
% a.out
Compiled by a FORTRAN compiler
%
If you like challenges, you can mark this article for later reference,
otherwise skip quickly to the supplied solution (hold on to your hat!).
% cat liar
program liar (*
c *) (output); (*
c *) begin
writeln (*
1 ( *,* ) 'Compiled by a FORTRAN compiler'
c *) (' Compiled by a Pascal compiler');
end . (*
c column 72 ^ *)
%
script done on Mon Jun 9 11:53:42 1986
Nice, no? The main tricks to notice are:
a) FORTRAN ignores all characters beyond the 72nd column (these were
originally used for sequence numbers).
b) '(*' and '*)' are Pascal's comment delimiters. In FORTRAN, the
character 'c' in the first column comments out the whole line,
while a '1' in column 6 indicates a continuation of the previous
statement.
c) 'write' is a valid FORTRAN statement, while 'writeln' is correct
Pascal.
d) Pascal requires a period after the 'end' keyword (which is found
at column 73!!!)
Also, notice the blanks around the logical unit and format specification:
there are necessary to avoid nested comments (for Pascal, of course). For
some FORTRAN compilers, *any* character in column 1 as the effect of
commenting out the line and therefore a simpler (ah!) solution would be:
program liar
(output); begin
writeln (*
1 ( *,* ) 'Compiled by a FORTRAN compiler' *)
(' Compiled by a Pascal compiler');
end .
A further challenge: write a similar program for C and Pascal.
ps.
Before any comments are made about the good government money I wasted doing
this silliness, I should point out that only 10 minutes were required to
write and debug the program.
--
--------------------------------------------------------------------------------
mpdevine%watdaisy@waterloo.csnet (519) 884-7123 Michel P. Devine
mpdevine%watdaisy%waterloo.csnet@csnet-relay.arpa CS Dept., U. Waterloo
{allegra,decvax,inhp4,utzoo}!watmath!watdaisy!mpdevine Waterloo, Ont. N2L3G1long@ittvax.ATC.ITT.UUCP (H. Morrow Long [CRT]) (06/16/86)
> > mpdevine%watdaisy%waterloo.csnet@csnet-relay.arpa CS Dept., U. Waterloo > A further challenge: write a similar program for C and Pascal. > I did something similar in my International Obfuscated C Code Contest entry for 1986. It can be compiled by both the pascal and C compiler on our 4.2bsd Vax. H. Morrow Long ---cut here------- (*FOO); /********************************************************** * * * 1986 Obfuscated C program entry - H. Morrow Long * * * * This program can be compiled by both the 4.2 BSD C and * * pascal (pc) compilers. 'pc' will complain a bit but * * will generate an executable. * * * * To run: * * Copy this file to both foo.c and foo.p * * cc -o c_foo foo.c * * pc -o p_foo foo.p * * * **********************************************************/ #define program void #define Main(a,b) main() #define procedure void #define begin { #define end } #define writeln puts #define foobar (FOO\ *) program Main(input,output) begin writeln("Hello World"); end -- H. Morrow Long Member Research Staff - Knowledge Based Systems ITT-ATC Advanced Technology Group 1 Research Drive Shelton, CT 06484 Phone #: (203)-929-7341 x. 634 path = {allegra bunker dcdvaxb dcdwest ucbvax!decvax milford mit-eddie psuvax1 qumix sii supai tmmnet yale}!ittatc!long
wesommer@mit-trillian.MIT.EDU (William Sommerfeld) (06/18/86)
This one works a little better; it doesn't complain when compiled by the
Pascal compiler (at least under 4.3BSD)
Tricks:
- { } is a comment in Pascal, statement delimeters in C.
- /*) and (*/ to toggle between 'C' and Pascal.
- The only reason to have the #define openbrace in there is so
that the Pascal compiler doesn't give you a warning about having an
open brace character in the middle of a comment.
Anyone want to work on some more obscure ones???
Bill Sommerfeld
ARPA: wesommer@athena.mit.edu
UUCP: mit-eddie!wesommer (forwards)
(*foo);
#define openbrace {
/*)
program main(output);
begin
writeln('I was compiled by a Pascal compiler.');
end.
{*/
main()
openbrace
puts("I was compiled by a C compiler");
}levy@ttrdc.UUCP (Daniel R. Levy) (06/21/86)
In article <722@mit-trillian.MIT.EDU>, wesommer@mit-trillian.MIT.EDU (William Sommerfeld) writes: >This one works a little better; it doesn't complain when compiled by the >Pascal compiler (at least under 4.3BSD) >Tricks: > - { } is a comment in Pascal, statement delimeters in C. > - /*) and (*/ to toggle between 'C' and Pascal. > - The only reason to have the #define openbrace in there is so >that the Pascal compiler doesn't give you a warning about having an >open brace character in the middle of a comment. >Anyone want to work on some more obscure ones??? > Bill Sommerfeld > ARPA: wesommer@athena.mit.edu > UUCP: mit-eddie!wesommer (forwards) >(*foo); >#define openbrace { >/*) >program main(output); >begin > writeln('I was compiled by a Pascal compiler.'); >end. >{*/ >main() >openbrace > puts("I was compiled by a C compiler"); >} Good show. Can anyone write one of these which combines Pascal, C, AND Fortran-77? I tried to do this but kept running into a situation where I needed a #define, which must be on the first character of a line (Fortran doesn't like that one bit since its comments are determined by the first character of each line). Oh yes, Fortran-77 considers '*', 'c', and 'C' all to be valid comment characters (begin a comment line by being in column 1), if that's any help (also a completely "blank" [columns 1-72] line is considered to be a comment). It's trivial to do a C/Fortran combination, however: main(){printf("I was compiled by a C compiler\n"); /* [starts in column 73] C ^ C | C Column 73 C | C v write(*,*)'I was compiled by a Fortran compiler' */ -- ------------------------------- Disclaimer: The views contained herein are | dan levy | yvel nad | my own and are not at all those of my em- | an engihacker @ | ployer or the administrator of any computer | at&t computer systems division | upon which I may hack. | skokie, illinois | -------------------------------- Path: ..!{akgua,homxb,ihnp4,ltuxa,mvuxa, vax135}!ttrdc!levy
dik@zuring.uucp (Dik T. Winter) (06/21/86)
In article <7772@watdaisy.UUCP> mpdevine@watdaisy.UUCP writes: > > During my CS340 lectures (a second course in Data Structures), I alluded to >some of the unusual features of some programming languages, in particular Pascal >and Fortran. On the spur of the moment (as an (spectacularly unsuccessful) >attempt at stimulating class participation), I suggested an exercise: write a >program which prints the identity of the compiler used to compile it. In the >following script session, both liar.f and liar.p are links to the data file liar >(which is the program in question). > Reminds me. Eons ago I wrote a program for the CDC Cyber which would be compiled by 4 compilers (Fortran, Pascal, Algol 60 and Algol 68). The most remarkable of the program was not that it passed 4 compilers, but it reproduced itself (regardless of the compiler, however, the Algol 60 compiler choked because of some compiler restrictions). This brings on another question. Is it possible to create non-trivial input to {t,d}itroff that reproduces itself? (By non-trivial I mean it should contain at least one line starting with period or apostrophe, because, as a silly co-worker pointed out to me, the empty file reproduces itself. Also .sy should not be permitted.) -- dik t. winter, cwi, amsterdam, nederland UUCP: {seismo,decvax,philabs,okstate,garfield}!mcvax!dik or: dik@mcvax.uucp ARPA: dik%mcvax.uucp@seismo.css.gov
jeff@pixar (Jeffrey Mock) (06/23/86)
Someone else try the harder case of using this clever `programming
methodology' to write a program that produces the text of the program
as output. This program will print itself regardless of being
compiled with a C or Pascal compiler. I just spent a few minutes
trying to do it and the solution isn't as simple as I first thought.
jeff mock
{ucbvax, sun}!pixar!jefflwall@sdcrdcf.UUCP (Larry Wall) (06/27/86)
In article <301@zuring.UUCP> dik@zuring.UUCP (Dik T. Winter) writes: > This brings on another question. Is it possible to create non-trivial > input to {t,d}itroff that reproduces itself? (By non-trivial I mean > it should contain at least one line starting with period or apostrophe, > because, as a silly co-worker pointed out to me, the empty file > reproduces itself. Also .sy should not be permitted.) Well... I took this as a challenge, and wrote something that may do what you want. I don't have {t,d}itroff real handy, but I did come up with a self-reproducing nroff script, so you might try it out with {t,d}itroff and see if it works. Not only was it NOT trivial, it was durn near impossible! I had to get rather devious. If there's an Obfuscated Nroff Code Contest in the near future, consider this my entry for it. Since there may be some people who wouldn't try doing this until they find out that it IS possible, I'll just say to them that it is. Have fun. So that I don't spoil their fun, the one I wrote follows both in uuencoded and rot13 format for those who aren't interested in trying to solve the problem themselves. I had another version that used '.eo' to reduce the number of escape characters, but it came out to the same number of bytes, and more lines, so I won't post it. If anyone can shorten this one, they're welcome to try. To run this, uudecode this_article nroff self.n >self.out cmp self.n self.out (By the way, I leave you with the challenge to make a C/nroff program that compiles without error and prints out "hello world\n" either way. Yes, it can be done. If anything it's easier than the self-reproducer, but that's not saying much. Definitely an AHA! type problem.) Larry Wall sdcrdcf!lwall THE UUENCODED VERSION: begin 664 self.n M+F5C.`HN96U:6@HN;G)Y>3$P,#`M,3(*+FYR>GHQ,#`M,0HN;F8*+F1E6%@* M+G1R.#AN*'EY"C@X(3@X(3@X)#$*+G1R.#AN*'IZ"BYA;5I:+BX*.#@A.#@A M+EA8.#@D,0HN;G)N;#`M,0HN+BX*+BX*+EA8+F5C.0HN6%@N96U:6@HN6%@N M;G)Y>3$P,#`M,3(*+EA8+FYR>GHQ,#`M,0HN6%@N;F8*+EA8+F1E6%@*+EA8 M+G1R.3EN*'EY"BY86#DY(3DY(3DY)#$*+EA8+G1R.3EN*'IZ"BY86"YA;5I: M+BX*+EA8.3DA.3DA+EA8.3DD,0HN6%@N;G)N;#`M,0HN6%@N+BX*+EA8+BX* ` end (***SPOILER WARNING***) Reading the ROT13 version may give you clues you don't want, despite the encryption. THE ROT13 VERSION: Abgr gung gur frpbaq unys vf vqragvpny gb gur svefg unys, jvgu gur rvtugf genafyngrq gb avarf naq n .KK cercraqrq. .rp8 .rzMM .aell1000-12 .aemm100-1 .as .qrKK .ge88a(ll 88!88!88$1 .ge88a(mm .nzMM.. 88!88!.KK88$1 .aeay0-1 ... .. .KK.rp9 .KK.rzMM .KK.aell1000-12 .KK.aemm100-1 .KK.as .KK.qrKK .KK.ge99a(ll .KK99!99!99$1 .KK.ge99a(mm .KK.nzMM.. .KK99!99!.KK99$1 .KK.aeay0-1 .KK... .KK..