Cody Boone Ferguson
55d09331d7
I have provided some examples as well and included how to use other diff programs like colordiff. Checking formatting (including how it looks on GitHub) and typo checks notwithstanding this entry is in good condition.
37 KiB
IOCCC news | People who have won | Winning entries
Known Bugs & (Mis)features | How to fix an IOCCC winner | Thanks for the fixes | A bit of web site history
The IOCCC FAQ
Q: How many entries do you receive each year?
By tradition, we do not say.
Q: How many judging rounds do you have?
Are you trying to trick us? We will not say that either.
Q: What are the general Makefile rules used in order to clean and build entries for use?
In general the best way to compile everything in an entry directory is to run:
make clobber all
If you wish to compile every entry for every year you should go to the top level
directory and run the same command as above, make clobber all.
The all rule will build everything necessary except for the alternate
versions. Every Makefile has an alt rule but it will only do something if an
alternate version exists. To build all the entries along with any alternate code
you can do from the top level directory:
make clobber everything
If you wish to only build alternate code you can do:
make clobber alt
If you wish to not remove the entry binaries and only compile the additional alt versions:
make alt
The following Makefile rules should be in all Makefiles:
- all: build the entry programs (main program and any supplementary program)
- alt: build alternate code
- clobber: clean up object files and all binary files (except for those that are not compiled)
- clean: a simpler version of
clobberthat only removes object files.make clobberdepends oncleanso runningmake clobberwill invokemake clean. - everything: equivalent to
make all alt.
Are there any other rules? You tell us!
Q: How come some entries have code that is incongruent with what the author(s) wrote about the entry?
It is very likely in this case that the code was fixed to work for modern
systems as part of the reworking of the website. If you have this problem in
some entries you should look at the original code as in winner.orig.c or
prog.orig.c. Sometimes the original is in an alt version like winner.alt.c
or prog.alt.c.
Q: Why have some entries that originally used gets() been modified to use fgets()? Doesn't this tamper with the entry too much?
A fine line indeed has to be drawn here but it was decided that it is worth it because of alarming warnings that can be displayed, in some systems at runtime interspersed with the output of the program.
For instance in macOS the entry 1990/tbr would output the warning in such a way that caused confusing output for the entry, looking like:
$ ./tbr
$ warning: this program uses gets(), which is unsafe.
# nothing here, what to do?
In some cases changing the code to use fgets() is not so easy to fix and in
one case at least there is an alternate version that has the fix instead due to
a problem it creates (correct output but segfaults after the output in one of
the forms of input).
In some cases it is not possible to fix or at least it is highly unlikely and so
those have mainly not been touched except one that has had the buffer size
increased (which could be done for others that are not possible to change to
fgets() though this has not been done).
Some entries can be made to look almost identical to the original entry. For
instance the fix to 1988/reddy required only a single
#define be added.
In the future we, the judges, would prefer that entries use fgets() to prevent
these problems.
NOTE: due to 'compatibility reasons' fgets() stores the newline and gets()
does not. We're not sure how this is compatibility but either way it can cause a
problem and it is this that has complicated most of the fixes though again some
can look almost identical.
Q: How can I easily see what was changed in order to get an entry to work in modern systems?
Although the thanks-for-fixes.md file sometimes gives commands to tell you how to do this, we have set up make rules to easily do this. For these you should be in the directory of the entry you wish to see the diff output.
If you want to see the difference from the original source try:
make orig_prog_diff
make: [Makefile:158: orig_prog_diff] Error 1 (ignored)
but this is perfectly fine and expected.
If however you wish to see the difference between the alt code and the entry itself, try:
make alt_prog_diff
NOTE: this might show at the end something like:
make: [Makefile:163: alt_prog_diff] Error 1 (ignored)
The alt_prog_diff rule will do nothing if no alt file exists.
If the alt code is the same as the original, say 1984/anonymous, there is no point in using this rule.
As some examples we'll first look at one, that has really long lines which
will make it harder to see what is different,
2001/anonymous. What you would do is cd 2001/anonymous and then do:
make orig_prog_diff
and then be really confused! :-)
But for an entry like 1991/dds, you can see the
differences much more easily. 1991/dds is a good example where it's very
simple to see what is different as it's just a couple lines.
You might be quite surprised how little some entries had to be changed and at the same time how much other entries had to be changed, often with quite complex differences! In some cases if the line is rather long, like the above mentioned one, it will be harder to see what changed but in other cases like 1984/decot or 1986/wall it's a lot easier.
Well, at least it's easier see the differences on a line-by-line basis but maybe not what actually changed, especially since it's easier to know what was fixed when you have compiler errors :-) (though there are, as noted, some examples where it's quite easy to see the differences).
1991/dds is also a good example to see the alt difference
very easily. To do that cd 1991/dds and then do:
make alt_prog_diff
and you'll see a single line changed and very simply.
Tip: if you have colordiff installed it's a lot easier to see the differences
To use these rules but provide a different diff, for instance colordiff,
just do:
make DIFF=colordiff orig_prog_diff # for original diff
make DIFF=colordiff alt_prog_diff # for alt diff
Q: I cannot get entry XYZZY from year 19xx to compile!
Some winners, particularly winners from long ago, no longer compile on more modern systems because the C language has evolved (i.e. the modern C compilers are much more strict in what they accept as a valid program), or the entry depended on operation system and library features that where common back then but are different/missing today.
E.g. it is highly unlikely that you'll be able to execute 1984/mullender on a computer you're using to read this FAQ. However for that entry Cody Boone Ferguson added alternative code so that you can enjoy this entry. Try:
make alt
./mullender.alt [microseconds]
The microseconds defaults to 10000.
Thank you Cody!
See also Yusuke Endoh's entry 2015/endoh3 which lets one compile it and run it. Another entry that you can enjoy it under is Christopher Mill's entry 2018/mills which is a PDP-7 emulator as well as a PDP-11/40 emulator.
Others are not so easy though we're working on this and over time have added alternative code and/or fixed them. Most entries do now work and the others we are working on (slowly as other things are also being done and this is on free time).
In some cases we replaced the original code with code that works for modern
systems but one can view the original code in the .orig.c files (sometimes the
original code is also in the directory as a winner.alt.c or prog.alt.c).
Some entries should not have modern system versions replaced. See below.
Q: I can't get some entries to work in 64-bit systems that don't support 32-bit!
Unfortunately some older entries are non-portable and require 32-bit support of 32-bit binaries. A problem system here is macOS Catalina (10.15) as as of that version macOS no longer supports 32-bit binaries.
There are numerous example entries that require 32-bit binaries. We have tried to note these in both the respective Makefiles and README.md files but it is possible that some were missed. These entries are very likely in the bugs.md file and we welcome any help in making an alternate version for 64-bit systems. Many were fixed to work with modern systems but some are supposed to only work with 32-bit systems so any updated version of these entries should be an alternate version.
Q: Under macOS I can't compile some entries and/or they don't work right. Why?
If the entry requires gcc and you did not explicitly install gcc in macOS you
will not be able to run or use these entries. This is because macOS gcc is
actually clang, even /usr/bin/gcc.
That being said many (if not most) of these entries have been fixed and some others will be looked at, when found.
Q: What is cb that is mentioned in some of the older entries?
This was a C beautifier for Unix, both AT&T and Berkeley, but it seems to no
longer be available, code wise, except for Plan 9, but Plan 9 was never used for
judging the IOCCC. A Unix man page for cb
still exists.
Q: I can't get XYZZY entry to compile with clang. What can I do?
Although we have fixed numerous entries to work with clang (sometimes in an alt version but usually in the program itself) there are some that simply cannot be fixed or if they are fixable they have not yet been fixed (we are working on this but other things have to be done too and all on free time).
This is because clang has some defects where the args of main() are required to be a specific type and some versions of clang allow only 1, 2 or 3 args, not 4, to main(). In the case of types of args many were changed to the right type and then what was main() became another function of the original main() type.
At the same time some entries are not designed to work with clang. There might be alternate code added at some point but as above this depends on free time and other things that have to be done plus remembering to do it.
Q: After running a program my terminal is all messed up! How do I restore my terminal?
The simplest way to do this is to type reset. If echo was disabled you can get
usually away with stty echo. Sometimes you can also get away with stty sane.
reset does the most but note that it will clear the screen (obviously clear
will too but it won't reset the terminal).
Q: How do I get X11 entries to work with macOS Mountain Lion and later?
In macOS Mountain Lion and beyond to run X11 applications one needs to install XQuartz. This will let you compile, link and run X11 applications.
Q: How do I compile and run entries that use SDL1/SDL2 ?
This depends on your operating system but below are instructions for linux and macOS with alternative methods for macOS and different package managers with linux.
Red Hat based linux
Execute one of the following as root or via sudo:
dnf install SDL2 SDL2-devel sdl12-compat sdl12-compat-devel
yum install SDL2 SDL2-devel sdl12-compat sdl12-compat-devel
In the past it was said that one might need to run make like:
make ... SDL2_INCLUDE_ROOT=/usr
or set the following environment variable:
export SDL2_INCLUDE_ROOT=/usr
but this might not be necessary in more modern days especially as we use
sdl-config and sdl2-config which should find the proper paths.
Debian based linux
Execute the following as root or via sudo:
apt install libsdl1.2debian libsdl1.2-dev libsdl2-dev
and then try make all again.
In the past it was said that one might need to run make like:
make ... SDL2_INCLUDE_ROOT=/usr
or set the following environment variable:
export SDL2_INCLUDE_ROOT=/usr
but this might not be necessary in more modern days especially as we use
sdl-config and sdl2-config which should find the proper paths.
Other linux distributions
Use your package manager to install the appropriate packages. Try the search feature of the package manager to determine which packages you need to install. Note that you might have to install both the library and the developmental packages: one for compiling and one for linking / running.
macOS
If you're using macOS there are at least three ways to obtain it. You can download it from the SDL website and install the package. That will not work well for the IOCCC but these will:
MacPorts
If you haven't already, install MacPorts. Then run:
sudo port install libsdl libsdl2
Homebrew
If you have not already done so, install Homebrew.
Then to install SDL and SDL2, execute the following command:
brew install sdl2 sdl12-compat
eval "$(/opt/homebrew/bin/brew shellenv)"
Q: How do I compile and run entries that use sound in macOS?
This might depend on the entry but in some cases like 2001/coupard one needs to do more work in order to get it to work. In this case you should be able to use the Swiss Army Knife of sound processing programs, SoX. To install this easily you can use either MacPorts or Homebrew. See below for instructions for each.
Usually the README.md file will explain how to use it in linux so we do not include this here, at least for now.
MacPorts
If you haven't already, install MacPorts. Then run:
sudo port install sox
Homebrew
If you have not already done so, install Homebrew.
Then to install SoX, execute the following command:
brew install sox
eval "$(/opt/homebrew/bin/brew shellenv)"
Q: How did entry XYZZY win? It breaks rule 2!
As entries have been fixed it is entirely possible that some of the entries no longer fit within the year's size restrictions. Invariably the length of columns and/or number of rows have also changed.
For the original version see the /archive directory where you can
find all the original winning entries. In some cases the winner.alt.c is the
original source code.
Q: Since some entries have been modified over time, how can I view the original entry?
See either the winner.orig.c or prog.orig.c, depending on the year (earlier
years we did not rename the code to prog.c but had it as the winner handle),
in the winning directories.
Q: I found a bug in a previous winner, what should I do?
We do not 'maintain' the contest winners as such. The code is made available on an 'AS IS' basis. If you have a FIX for an entry, we suggest that you try and email it to the authors and to submit your fixes in a GitHub pull request (with ONE PULL REQUEST PER FIX, please!).
Note that just because you have a fix does not mean it'll be accepted. This might be because the author objects or it doesn't fit in some way or another. Of course if the entry does not work we'll certainly be more inclined to accept the fix. If it is accepted we'll be happy to credit you in the entry's README.md file. If you're a previous winner we will happily link to your entries; if you're not we can link to your website if you wish.
More generally please see the file how-to-bugfix.md. Note that this file is not a tutorial on how to fix X, Y and Z problems but rather what to do to get the fix in.
Q: Do you have a list of entries with known problems and/or features that might appear to be bugs but are not?
Yes! Please see bugs.md for a list of known bugs and/or issues of a variety of kinds.
Note that just because an entry is not in the bugs file does not mean there is not an issue and note that some issues are simply missing files, dead URL(s) or something like that.
Q: Are there types of entries that are submitted so frequently that the judges get tired of them?
Yes, there are types of entries that show up over and over again.
The guidelines say:
We tend to dislike programs that: are similar to previous winners
We like variety. However too often we see (please look at the winning examples given to be aware of the level of the competition):
maze generator
tic-tac-toe game
solitaire/Othello game
generating small primes (below is the list of all prime related winners)
As you can see, just a list of primes (let alone small primes) does not cut it anymore.
self reproducing program
-
1994/smr - _do not claim your program is the smallest one without seeing this!
-
2000/dhyang - unless you beat this one, your chances are slim
entries that just print "Hello, world!"
...it's so 20th century...
entries that use some complex state machine/table to print something
rot13
pi or e computation
Hints on overused themes
HINT: It is not fatal to send in those types of entries, it is just MUCH HARDER to win. Such an entry would have to do something really unique AND interesting to even make it into the final judging rounds. Entries that reuse these 'worn out' themes frequently lose to other entries that do something different.
HINT: If you really MUST send in an entry based on an 'overused theme', be sure that it is obfuscated in several new and novel ways. Be sure to clearly explain near the beginning of your remarks why you are submitting a entry based on an 'overused theme' and why the judges should not simply toss it out as being boring.
Q: What should I write in the 'remarks' (remarks.md) section of my entry, if anything at all?
As much or as little as you wish.
What helps:
-
explaining what your entry does
-
how to entice it to do what it is supposed to do
-
what obfuscations are used
-
what are the limitations of your entry in respect of portability and/or input data
-
how it works (if you are really condescending)
What does not help:
-
admitting that your entry is not very obfuscated (you see, the contest is called the IOCCC, not the INVOCCC :-) ); but even if you do not admit it, not very obfuscated entries have a minuscule chance to win (although 2000/tomx is a notable counterexample).
-
mentioning your name or any identifying information in the remark section (or in the C code for that matter) - we like to be unbiased during the judging rounds; we look at the author name only if an entry wins. See the guidelines if this is not clear!
-
leaving the remark section empty.
Q: I am confused: what should I put to the 'build' section?
The 'build' section must contain UNIX shell-style commands to compile your C language entry in a file named prog.c and produce an executable file. More than one command is allowed, as far as the size rule is followed. Usually one (or more) of these commands will be an invocation of a C compiler (use cc or gcc), but there may be exceptions (or not anymore, see 2000/tomx).
If your entry depends on a particular source file name during compilation or execution, you will have to put a line
cp prog.c desired_name.c
before the compiler invocation line.
If in doubt, put
cc -o prog prog.c
in the 'build' section.
Note however that with the mkiocccentry tool we have an example Makefile which you should update and submit, using the instructions above if necessary.
Q: What happened to the winners for 1997, 1999 2002, 2003, 2007 - 2010, 2016, 2017, 2021 and 2022 ?
There are none. There was no IOCCC in those years.
Q: What are the .orig.c files in the directories in winning entries ?
Due to the fact that the original code has sometimes had to change these files are the original winning entry or as close to as possible to the original that we can find.
Q: Why don't you publish non-winners?
Because the publication on the IOCCC site IS the award! Anyone is free to put their IOCCC hopefuls, lookalikes and/or non-winners on their web page for everyone to see.
Q: How much time does it take to judge the contest?
It takes a fair amount of time to setup, run, answer email, process entries, review entries, trim down the set entries to a set of winners, doing the write-up of the winners, announcing the winners, reviewing final edits of the winning entry set, posting the winners, being flamed :-), tell folks who send in late entries to wait until the next contest, etc... It takes a few weekends and a number nights of study and work ... which is hard given that we are busy with many other activities as well.
Note that we do not contact the author if an entry does not compile or does not work as advertised, we might attempt to fix obvious compilation problems or incompatibilities, but no more than that - so be sure that your entry does work on at least a couple different platforms, at least one of them being UNIX or POSIX-conforming.
Q: What platform should I assume for my entry?
Your entry must compile with GCC and run under at least one flavor of UNIX (e.g. Linux or Solaris). To improve chances to win, it should work under both BSD and SysV style UNIX (think stty cbreak), and should not depend on endianness. If it can compile and run on Windows and/or Mac (see 2000/thadgavin for one example, amongst others), even better. Being able to compile with other compilers like clang is also a good thing.
Q: I would like to mirror the IOCCC web site. May I do so?
We are not accepting mirror requests at this time, sorry.
Q: I want to publish some parts of the IOCCC in an article, or book, or newsletter, or use then in class/instructional notes, or quote from the IOCCC. May I do so?
Please ask the IOCCC judges first. Please send your request using the instructions on the contacting the IOCCC Judges page.
Q: What are the grand prize / Best of Show winners?
In 1984-1987, the grand prize winners are:
Starting from 1988, the entry we liked the most in that year is called "Best of Show". Here are the "Best of Show" entries:
In 1993, 1994 and 1995 the judges were unable to select a clear overall winner. So to give a nod to the entry that had the highest approval ranking from the judges, they used the following awards:
-
1993/rince - Most Well Rounded
-
1994/shapiro - Most Well Rounded
-
1995/leo - Best Use of Obfuscation
These could be considered the 'best entry' for those years with 1 or more other entries that came in close behind.
Q: I managed to get entry XYZZY from year 19xx to compile; now it fails to run!
switch(19xx/XYZZY) {
case [1984/mullender](years.html#1984_mullender):
try the alternative code added by Cody Boone Ferguson like:
make alt
./mullender.alt [microseconds]
where microseconds is the number of microseconds to sleep between writes. Default 10000. This is useful to see different speeds as well as if your CPU is too slow or too fast for the entry. See the README.md for more details.
Thank you Cody for your help!
BTW: in this case it's a slight fix and improvement to our original suggested code:
main(){int i=512;do write(1," :-)\b\b\b\b",9),usleep(i);while(--i);}
BTW: is there such a thing as too fast a CPU ? :-) actually yes for certain code which is probably not as uncommon as you think :-) ).
break;
- case 1991/davidguy: disable access control of your X server (xhost +). If it does not help (or if you see black screen), restart the X server without access control and with visual depth of 8 bits or less.
break;
- default: Try enabling
-fwritable-stringsfor GCC or its equivalent for other compilers.
}
Q: How did the IOCCC get started?
One day (23 March 1984 to be exact), Larry Bassel and I (Landon Curt Noll) were working for National Semiconductor's Genix porting group, we were both in our offices trying to fix some very broken code.
Larry had been trying to fix a bug in the classic Bourne shell (C code #defined to death to sort of look like Algol) and I had been working on the finger program from early BSD (a bug ridden finger implementation to be sure).
We happened to both wander (at the same time) out to the hallway in Building 7C to clear our heads.
We began to compare notes: 'You won't believe the code I am trying to fix'.
And: 'Well you cannot imagine the brain damage level of the code I'm trying to fix'.
As well as: 'It's more than bad code, the author really had to try to make it this bad!
After a few minutes we wandered back into my office where I posted a flame to net.lang.c inviting people to try and out obfuscate the UN*X source code we had just been working on.
BTW: I (Landon Curt Noll) had to post this typo correction. Thus began the tradition of putting typos in the contest rules and guidelines ... to make them more obfuscated of course! :-)
BTW: This posting was made back in the days when AT&T was the evil giant. Now, Microsoft makes AT&T look mild and kind in comparison. :-( (IMHO) ).
BTW: See the story about the 'Bill Gates' award. :-)
OK, back to the story.
We (Larry and I) received a number of entries by email. When we began to receive messages from outside of the US, Larry and I decided to include International in the name.
The 1st IOCCC winners were posted on 17 April 1984.
There were 4 winners in 1984:
BTW: The (dis)honorable mention wished to remain anonymous. While many have asked who it was, we have continued to follow the author's wish to remain anonymous.
A few years ago, we asked the author if they still wanted to remain anonymous. They said: 'Yes, I want to keep my anonymity. But you can tell them that I am well known for my connection to the C language'. It was not until 2001 that another anonymous entry received an award.
BTW: The 1984 winner remains one of my (Landon Curt Noll) all time favorites.
The name used in the posting of the 1st IOCCC winner posting was International Obfuscated C Code Contest or IOCCC for short.
The posting said '1st annual', so in 1985 we held the 2nd IOCCC contest and the tradition continues as the longest running contest on the Internet.
P.S. Part of the inspiration for making the IOCCC a contest goes to the Bulwer-Lytton fiction contest.
P^2.S. See the overall README for more details.
 |
© Copyright 1984-2020,
[Leo Broukhis, Landon Curt Noll](/judges.html)
- All rights reserved |
Q: Why do you sometimes use the first person plural?
As a precedent for first person plural, we may sight Two-, Three-, and Four-Atom Exchange Effects in bcc in 3He and the co-authorship of F. D. C. Willard as well as the APS New Open Access Initiative.
The number of IOCCC judges has always been "> 1" such that IOCCC judges often prefer to themselves in the plural, sometimes in the common plural, sometimes in the first person plural, there may be some who reject such a practice. Therefore please consider that when you believe that just one of the IOCCC judges is in action, you may consider that F. D. C. Willard has been consulted, agreed, and has authorized the statement thus allowing the in the common plural to be used without offense.
And while some might claim that F. D. C. Willard died in 1982, we suggest that the Schrödinger's cat superposition may still be in effect and the 1982 report of death was an exaggeration.
p.s. Here is an image of F. D. C. Willard:
Q: Why do Makefiles use -Weverything with clang? Don't you know that its use is not recommended by clang developers?
The use of -Weverything is limited to when one forces CC=clang. Users with
clang compilers are not required to set CC=clang but when they do,
-Weverything is enabled with all of its challenges, pedantic warnings, and
sometimes warnings about things that do not matter, some of which are frankly
frivolous and often downright dubious.
To enable this feature:
make clobber all CC=clang
or:
make clobber all 'CWARN+= -Weverything'
though it should be noted that if one tries -Weverything with compilers that
are not clang they might see something like:
echo 'int main(void) {}' > foo.c ; cc -Weverything foo.c -o foo
cc: error: unrecognized command-line option '-Weverything'
which means that it can't even be compiled. Thus the proper way to do it is the first one.
IOCCC authors who have access to a clang compiler might wish to try they their
hand at compiling with -Weverything while using a minimum of -Wno-foo
statements. Sometimes there is a technical or pedantic issue that
-Weverything warns about that would merit a change to your C code. Of course
if you're running out of bytes due to rule 2[ab] one might not have much choice.
Thus is something that obfuscators simply sometimes have to deal with!
If you to try to use minimize the number of -Wno-foo options needed with
-Weverything, please mention this in your remarks about the entry, as the
judges note you attempt to honor it (see also below). In some cases your
obfuscated code will issue warnings with -Weverything no matter what: the
-Wno-poison-system-directories is a common example of this but there are
others as well.
If you do try for a warning clean -Weverything, keep on mind that while your
compile environment might be warning free, a different clang version or a
different build environment might still have warnings. For instance the warning
set is different in macOS (which by default is clang even when run as gcc!)
than linux! Given that your entry MUST work as documented, you may be safer to
say that your entry keeps the number of warnings and -Wno-foo options while
compiling with clang -Weverything at a minimum. Because if you claim zero
warnings, and we find a warning situation, this may diminish the value of your
entry as it is not as documented. Thus it might be wise to point this out and
also if you can test it in multiple platforms (or versions of clang, see
below note) this would be advisable.
NOTE: different versions of clang have other differences as well. For instance
a defect of clang that required numerous entries to be fixed is that it
requires that main()'s arguments to be of a specific type. However some
versions of clang are more strict in the number of args allowed. These reasons
are part of why numerous entries had to be modified so that main() calls
another function instead of doing it all in main(). Another reason was that
some entries that recursively called main() caused a crash or otherwise broke
the entry in modern systems. Some entries do not work in clang (or at least do
not work completely) due to these defects, for instance
1989/westley.
As you can see, using clang has some additional problems to work out but if
you can get your entry to work well in clang it might very well be considered
better than other entries.