Updates and fixes in 1991/brnstnd

Make sorta.README a markdown file (sorta.README.md) and update
formatting for it.

Typo fixes and formatting fixes in README.md.
This commit is contained in:
Cody Boone Ferguson committed 2023-04-09 08:08:38 -07:00
1 parent ececfced3e
commit 68c57a4bb1
2 files changed
+128 -129

No files matched your search

+88 -85
View File
@@ -1,11 +1,13 @@
# Best Of Show:
Daniel J. Bernstein
New York University
Daniel J. Bernstein
New York University
## To build:
make all
```sh
make all
```
[Cody Boone Ferguson](/winners.html#Cody_Boone_Ferguson) fixed this for modern
systems. There were two invalid operands to binary expression (`char *` and
@@ -18,20 +20,22 @@ were resolved with the comma operator. Thank you Cody for your assistance!
## Try:
./brnstnd < sorta.i2+2
./brnstnd < sorta.iwhosort
```sh
./brnstnd < sorta.i2+2
./brnstnd < sorta.iwhosort
```
See also the other `sorta.*` files.
## Judges' comments
## Judges' comments:
For info on more examples, read the sorta.README file.
For info on more examples, read the [sorta.README.md](sorta.README.md) file.
The author wished to win the "most useful program" award and
documented this in the source code. The judges were unmoved
by this blatant attempt to influence the contest and rejected
this idea... so we gave it the Best of Show award instead!!
The author wished to win the "most useful program" award and documented this in
the source code. The judges were unmoved by this blatant attempt to influence
the contest and rejected this idea... so we gave it the `Best of Show`
instead!!
NOTE: One should remove the final trailing newline to obtain the
original source file. This step is not needed to compile
@@ -40,7 +44,7 @@ this entry.
To have a chance to compile under a modern CPP, we had to
replace `#D` with `#define`.
## Author's comments
## Author's comments:
This is an interpreter for the programming language SORTA, a systems
and numerical programming language with features sorta from C, sorta
@@ -59,104 +63,103 @@ instance, not to mention infinite-depth tail recursion.
SORTA ignores its arguments (though it makes them available to the
script); it takes all commands, character by character, from its
standard input. Unrecognized commands are repeated with a ?.
standard input. Unrecognized commands are repeated with a '?'.
SORTA maintains an ``i stack'' for integers and an ``s stack'' for
strings. It also keeps track of ``programs,'' one for each character.
Operations are silently ignored if the relevant stacks are too low,
except as noted for s, S, and !. If the i stack or s stack (or the stack
of buffered macro commands) grows too high, SORTA will exit silently
with exit code 2. (Compile parameter -Do=250 controls what ``too high''
means; you should not make o larger than 250.) I don't think it's
possible to crash the interpreter.
SORTA maintains an `i stack` for integers and an `s stack` for strings. It also
keeps track of `programs`, one for each character. Operations are silently
ignored if the relevant stacks are too low, except as noted for `s`, `S`, and
`!`. If the `i stack` or `s stack` (or the stack of buffered macro commands)
grows too high, SORTA will exit silently with exit code 2. (Compile parameter
`-Do=250` controls what 'too high' means; you should not make `o` larger than
250.) I don't think it's possible to crash the interpreter.
Here, then, is the SORTA programming language. All non-digits delimit
numeric constants. Spaces and newlines are ignored except as numeric
delimiters.
Basic operations:
### Basic operations:
q: quit
number: push that number on top of the i stack
"string": push that string on top of the s stack, no length limit
[string]: push that string on top of the s stack, no length limit
#: make top of i stack non-destructively in ASCII, push result onto s stack
`: print top of s stack non-destructively
$: print top of s stack non-destructively without newline
d: drop (pop) top of i stack
D: dup (duplicate) top of i stack
': dup top of s stack
s: pop top of i stack. If it is n, swap (n+1)th-top of i stack with top.
1s, for example, swaps the top two elements; 2s swaps the top with
the third down; etc. This always pops the top of the i stack, even
upon failure.
S: pop top of i stack, then act just like s but upon s stack
l: pop top of s stack, push its length onto i stack
a: push argc onto i stack
A: pop top of i stack, push argv[i] on top of s stack
T: concatenate top two elements of s stack
_: negate top of i stack
+: pop top two elements of i stack, add, push back sum
*: pop top two elements of i stack, multiply, push back product
/: pop top two elements of i stack, divide, push back quotient
>: pop top two elements of i stack, compare, push back 1 or 0 as in C
&: pop top two elements of i stack, nand, push back bitwise result
Note that nand is sufficient to construct all bitwise operations,
as demonstrated by icalc below.
- `q`: quit
- `number`: push that number on top of the i stack
- `"string"`: push that string on top of the s stack, no length limit
- `[string]`: push that string on top of the s stack, no length limit
- `#`: make top of i stack non-destructively in ASCII, push result onto s stack
- ` ` `: print top of s stack non-destructively
- `$`: print top of s stack non-destructively without newline
- `d`: drop (pop) top of i stack
- `D`: dup (duplicate) top of i stack
- `'`: dup top of s stack
- `s`: pop top of i stack. If it is `n`, swap `(n+1)`th-top of `i` stack with top.
`1s`, for example, swaps the top two elements; `2s` swaps the top with the
third down; etc. This always pops the top of the `i stack`, even upon
failure.
- `S`: pop top of `i stack`, then act just like `s` but upon `s stack`
- `l`: pop top of `s stack`, push its length onto `i stack`
- `a`: push argc onto i stack
- `A`: pop top of `i stack`, push `argv[i]` on top of `s stack`
- `T`: concatenate top two elements of `s stack`
- ` _ `: negate top of i stack
- `+`: pop top two elements of i stack, add, push back sum
- ` * `: pop top two elements of i stack, multiply, push back product
- `/`: pop top two elements of i stack, divide, push back quotient
- `>`: pop top two elements of i stack, compare, push back 1 or 0 as in C
- `&`: pop top two elements of i stack, nand, push back bitwise result
Note that nand is sufficient to construct all bitwise operations, as
demonstrated by icalc below.
System operations:
### System operations:
o: pop top of s stack and top two of i stack; open(s1,i2,i1); leave
result on i stack
O: pop top of i stack, close() it
u: pop top of i stack, dup() it, push result back on
F: fork(), push result on i stack. This is not always safe while SORTA
is reading keyboard input or a script, as the forked programs share
file descriptors. It is always safe inside a program (see below).
P: pipe(), push two ends <p0> <p1> onto i stack. In case of trouble,
push <?> <-1> onto i stack.
w: wait(), push successful result (or -1 if no children) on i stack
!: execvp() top of the s stack. Arguments are the next things down on
the s stack, in reverse order from popping; the first character of
each string is lopped off. There must be an empty string somewhere
in the s stack to terminate the argument list. If the execve()
fails, any operation involving current members of the s stack has
undefined effects, and -1 is pushed onto the i stack.
- `o`: pop top of `s stack` and top two of `i stack`; `open(s1,i2,i1);` leave
result on `i stack`
- `O`: pop top of `i stack`, `close()` it
- `u`: pop top of `i stack`, `dup()` it, push result back on
- `F`: `fork()`, push result on `i stack`. This is not always safe while SORTA is
reading keyboard input or a script, as the forked programs share file
descriptors. It is always safe inside a program (see below).
- `P`: `pipe()`, push two ends `<p0> <p1>` onto `i stack`. In case of trouble,
`push <?> <-1>` onto `i stack`.
- `w`: `wait()`, push successful result (or -1 if no children) on `i stack`
- `!`: `execvp()` top of the `s stack`. Arguments are the next things down on
the s stack, in reverse order from popping; the first character of
each string is lopped off. There must be an empty string somewhere
in the s stack to terminate the argument list. If the execve()
fails, any operation involving current members of the s stack has
undefined effects, and -1 is pushed onto the i stack.
High-level language operations:
### High-level language operations:
:x: copy the top of the s stack into a program labeled by character x
=x: pop the top of the i stack. If it was nonzero, execute the program
labeled by character x. Note that a digit at the end of a program
will merge with any digits after =x; in that case you usually want
to add a space at the end of the program.
- :x: copy the top of the s stack into a program labeled by character x
- =x: pop the top of the i stack. If it was nonzero, execute the program
labeled by character x. Note that a digit at the end of a program
will merge with any digits after =x; in that case you usually want
to add a space at the end of the program.
Common idioms: To drop the top of the s stack, ld. To do a <, 1s>. To
unconditionally execute the program labeled by character x, 1=x (or any
nonzero number followed by =x). To print the top of the i stack
non-destructively, #` (or #$ if you don't want the newline). To
subtract, _+. To do mod or and or any of the other missing operations,
non-destructively, `# ` ` (or `#$` if you don't want the newline). To
subtract, `_+`. To do mod or and or any of the other missing operations,
combine the available operations as illustrated below. To introduce a
comment, [ this is a comment string which is promptly eliminated ]ld.
comment, `[ this is a comment string which is promptly eliminated ]ld`.
SORTA's requirements: it wants fork(), execvp(), open(),
close(), dup(), pipe(), and wait(), so it obviously won't even compile
SORTA's requirements: it wants `fork()`, `execvp()`, `open()`,
`close()`, `dup()`, `pipe()`, and `wait()`, so it obviously won't even compile
on a non-UNIX machine. It also assumes that you have a size-256
character set and that the characters between '0' and '9' are exactly
character set and that the characters between `'0'` and `'9'` are exactly
the digits in order. It does *not* depend on ASCII, despite the code
appearance. Also note that SORTA does not attempt to declare malloc(),
appearance. Also note that SORTA does not attempt to declare `malloc()`,
so you will get some warnings about illegal pointer combinations. Also
note that there are unreached statements in SORTA.
note that there are unreachable statements in SORTA.
While the program source of course depends highly on #defines for
While the program source of course depends highly on `#define`s for
obfuscation, I like to think that this code has those little touches,
those professionally sharpened edges that mark true software
engineering. Observe, for instance, how i and s are the string and
engineering. Observe, for instance, how `i` and `s` are the string and
integer stacks respectively. It's these little things that make you
feel at home in an otherwise utterly useless piece of code. They're
what make an obfuscation work.
The character pool (``I'' after cpp) makes it rather painful to
The character pool (`I` after cpp) makes it rather painful to
see the effect of commands at a glance. I can just imagine people
spending hours bouncing between the pool and the rest of the code,
or accidentally changing the pool without realizing its importance.
@@ -173,8 +176,8 @@ everywhere will make people think of ASCII, even though the code is
not ASCII-dependent. What's your first thought when you see
several arrays[256]? This is pretty standard obfuscation otherwise.
I like the way that unbalanced macro braces can throw off the
reader in ``if(c>0){y+1]=z[c];'', and I had fun covering the
alphabet with #defines, but that's nothing special.
reader in `if(c>0){y+1]=z[c];`, and I had fun covering the
alphabet with `#define`s, but that's nothing special.
Possible future extensions to SORTA include string extraction and
matching, reading from files into strings, and encrypting the string