Fix 2004/kopczynski and format fix README.md

The fix is that we cannot pass -ON to the compiler. With -ON it enters
an infinite loop as the variable l goes up, not down.

Format fix README.md.

Add kopczynski-a to show what happens when it's not a number.

2004/kopczynski/Makefile

@@ -66,7 +66,7 @@ CINCLUDE=
 
 # Optimization
 #
-OPT= -O3
+OPT= 
 
 # Default flags for ANSI C compilation
 #

2004/kopczynski/README.md

@@ -1,12 +1,8 @@
 # Best one-liner
 
-    Eryk Kopczynski
+Eryk Kopczynski
-    Faculty of Mathematics, Informatics and Mechanics
+Poland
-    ul. Banacha 2
+<http://www.mimuw.edu.pl/~erykk/>
-    02-097 Warszawa
-    Poland
-
-    http://www.mimuw.edu.pl/~erykk/
 
 ## To build:
 
@@ -20,6 +16,11 @@ make
 ./kopczynski < input_file
 ```
 
+[Cody Boone Ferguson](/winners.html#Cody_Boone_Ferguson) reported that this
+entry cannot be optimised by the compiler if it's to work. He also added the
+file [kopczynski-a](kopczynski-a) to demonstrate what happens when art more like
+a letter is fed to the program. Thank you Cody!
+
 ## Try:
 
 ```sh
@@ -31,6 +32,14 @@ make
 ./kopczynski < kopczynski-11
 ```
 
+Also try:
+
+```sh
+./kopczynski < kopczynski-a
+```
+
+What happens?
+
 ### Alternate code
 
 The judges modified the program to print its result.  The original
@@ -47,7 +56,7 @@ Then try:
 echo $?
 ```
 
-## Judges' remarks
+## Judges' remarks:
 
 What is in a line?  A lot when you obfuscate the way Eryk Kopczynski
 did it.  This small one line program is an outstanding technical
@@ -69,10 +78,10 @@ Here is a related puzzle: how to continue the sequence:
 
     1, 0, 0, 0, 0, 0, 1, 0, 2, 1, ...
 
-With certain arguments, you can alter the exit value.  Can you figure
+With certain arguments, you can alter the exit value / output.  Can you figure
 out how and why?
 
-## Author's remarks
+## Author's remarks:
 
 The program proves it is not as hard to recognize numbers. The current
 one liner version should correctly recognize all numbers from 8 to 11.
@@ -81,112 +90,108 @@ recognize negatives, zeros which have dots or slashes inside them,
 digits should be separated, and numbers should be "complete".)
 
 This program returns the number as its exit status. (In most shells,
-you can see the exit status of the last command with 'echo $?' (or
+you can see the exit status of the last command with `echo $?` (or
-just '$?').)
+just `$?`).)
 
-    [[NOTE: The judges altered the primary version to output a number.]]
+**NOTE: The judges altered the primary version to output a number.**
 
 On the input, give numbers formed as a sequences of empty spaces and #
 signs (you can also use dots instead of spaces), ending with an empty
 line. (No number has an empty line in it.) The program should return
 the number from 8 to 11 that it thinks it is.
 
-* The line is 131 bytes long (in 1986 165 bytes was OK, and the size
+* The line is 131 bytes long (in 1986, 165 bytes was OK, and the size
   limit has grown from 1986...).
 
 * I have decided that it is better for this program to return the
   number, not to print it, but if you think otherwise, the change would
-  cost 6 chars (plus a warning). Change 'return *' to 'printf("%d",*)'.
+  cost 6 chars (plus a warning). Change `return *` to `printf("%d",*)`.
-  ('$?' is shorter)
+  (`$?` is shorter)
 
-* I achieved some additional tightness by '-Dif=while', saved 9 chars
+* I achieved some additional tightness by `-Dif=while` which saved 9 chars of
-  of source for cost of 11 chars of build file. If I used '-DB=while',
+source for the cost of 11 chars of the [Makefile](Makefile). If I used
-  I would save 12 for 10, but I think the current version is slightly
+`-DB=while`, I would save 12 for 10, but I think the current version is slightly
-  more obfuscated.
+more obfuscated.
 
-* A remark on portability:
+### Portability
 
-    It only requires two's compliment arithmetics, as far as I
+It only requires two's complement arithmetics, as far as I know. The maximum
-    know. The maximum length of an input line is N-9, where
+length of an input line is `N-9`, where `N` is the size of `int` in bits
-    N is the size of int in bits (you can try to compile with
+(you can try to compile with an option `'-Dint=long long'` for longer
-    an option '-Dint=long long' for longer lines).
+lines).
 
-   -- Algorithm & Obfuscation --
+### Algorithm & Obfuscation
 
-1. This program works by calculating the Euler characteristics of input,
+1. This program works by calculating the [Euler
-   plus 9. It is the number of digits (i.e. separated #-regions) minus the
+characteristics](https://en.wikipedia.org/wiki/Euler_characteristic) of input,
-   number of holes in them. It turns out that this algorithm works
+plus 9. It is the number of digits (i.e. separated #-regions) minus the number
-   correctly for all numbers from 8 to 11.
+of holes in them. It turns out that this algorithm works correctly for all
+numbers from 8 to 11.
 
-2. We know that Euler characteristics is the number of "vertexes"
+2. We know that [Euler
+characteristics](https://en.wikipedia.org/wiki/Euler_characteristic) is the number of "vertexes"
 
-   #
+	    #
 
    minus the number of "edges"
 
-   ## # #. .#
+	   ## # #. .#
-      # .# #.   ('.' must be empty)
+	      # .# #.   ('.' must be empty)
 
    plus the number of "faces"
 
-   ##
+	   ##
-   ##
+	   ##
 
 3. If we fill all these patterns to 2x2 squares,
 
-   #? ## #? #. .# ##
+	   #? ## #? #. .# ##
-   ?? ?? #? .# #. ##
+	   ?? ?? #? .# #. ##
-   +1 -1 -1 -1 -1 +1
+	   +1 -1 -1 -1 -1 +1
 
-   substitute all possible "values" of '?' and sum up, we get that it is
+   substitute all possible "values" of `?` and sum them up, we get that it is
    enough to calculate:
 
-   #. .# ##
+	   #. .# ##
-   .. #. #.
+	   .. #. #.
-   +1 -1 -1
+	   +1 -1 -1
 
    Note now it is simple and obfuscated enough to be used as
    a "magic trick".
 
 4. I make a 16-character array; when indexed by 4-bit value which
    represents the 2x2 squares, it will return a number to add/subtract
-   {-1,0,+1}
+   `{-1,0,+1}`
 
-5. We convert lines of input to bitmaps (N-bit integers). Additionally,
+5. We convert lines of input to bitmaps (`N-bit` integers). Additionally, it is
-   it is heavily obfuscated by using negative numbers (two's compliment).
+heavily obfuscated by using negative numbers (two's complement).  For example,
-   For example, it turns out that 6 in "z=6" in the only 1-digit number
+it turns out that `6` in `z=6` is the only 1-digit number which works here (and
-   which works here (and it's not only because we have "z>5" in the
+it's not only because we have `z>5` in the following `while`). The conversion
-   following while). The conversion routine also sets n=8 in an obfuscated
+routine also sets `n=8` in an obfuscated and concise way.
-   and concise way.
 
-5. Since we always return the number from 8 to 11, we can calculate
+6. Since we always return the number from 8 to 11, we can calculate modulo 4.
-   modulo 4. Instead of 16 {-1,0,+1} values, we use a 15-character string
+Instead of `16 {-1,0,+1}` values, we use a 15-character string which agrees
-   which agrees modulo 4 to the array in (4). (NUL fits very well as the
+modulo 4 to the array in (`4`). (`NUL` fits very well as the 16th character, so
-   16th character, so we use 15-character string.)
+we use 15-character string.)
 
-6. Since the rule found in (3) did not produce any funny messages, I
+7. Since the rule found in (3) did not produce any funny messages, I changed it until it did. (The `#` character to find vertexes is in the top right corner, not the top left one.) The following messages will also work instead of `has dirtiest IF`:
-   changed it until it did. (The # character to find vertexes is in the
-   top right corner, not the top left one.) The following messages will
-   also work instead of "has dirtiest IF":
 
-   HighlandYachter
+	   HighlandYachter
-   TestDirtiestTen
+	   TestDirtiestTen
-   has lintiest IF
+	   has lintiest IF
-   TighterLaughter
+	   TighterLaughter
-   HustlerDaughter
+	   HustlerDaughter
-   TighterDemoThen
+	   TighterDemoThen
 
    They can be mixed, of course!
 
-7. [[NOTE: Item 7 was removed by the judges so as to not spoil the
+8. NOTE: item 8 was removed by the judges so as to not spoil the challenge of
-	 challenge of figuring how arguments can be used to alter
+figuring how arguments can be used to alter the exit value.
-	 the exit value.]]
 
-8. The variable names are confusing.
+9. The variable names are confusing.
 
-9. Finally, the program tries to be as concise as possible. Some
+10. Finally, the program tries to be as concise as possible. Some obfuscation
-   obfuscation comes from this. For additional obfuscation and
+comes from this. For additional obfuscation and conciseness, `if` means `while`.
-   conciseness, 'if' means 'while'.
 
 ## Copyright and CC BY-SA 4.0 License:
 

2004/kopczynski/kopczynski-a

@@ -0,0 +1,8 @@
+      ####
+     ######
+    ##    ##
+   ###    ###
+   ##########
+   ##      ##
+   ##      ##
+