Compatibility Config

Compatibility Config

These configuration options change how existing code evaluates. All of them are read at evaluation time, so changing one takes effect on the next call rather than at the next restart. Set them with @config/set or on the web portal’s configuration page.

Boolean compatibility: tiny_booleans

Boolean conditions use the same rules in functions, command guards, list filters, and indirect calls. With tiny_booleans off, numeric zero (including -0, 0.0 and hexadecimal zero), an empty string, space-only text, and values beginning #- are false. Other text is true, including the word false. A preserved tab is text, and a trailing space makes an otherwise numeric value text.

With tiny_booleans on, a leading signed integer determines truth. Thus text and 0.1 are false, while 1.2text is true. The compatibility conversion uses a signed 64-bit prefix, saturates overflow, then takes its low 32 bits. For example, 4294967296 is false.

> think t(-0.1)
1
> think t(-0)
0

neq() is the numeric inverse of eq(): it is true when not all arguments are equal. condall() returns every result with a true condition; ncond() and ncondall() select false conditions. A condition is a single value, not a list of separate booleans. A selected empty result does not cause the default to run.

Numeric compatibility: tiny_math, null_eq_zero

Arithmetic and numeric comparisons read tiny_math and null_eq_zero at call time, including indirect #apply calls. With tiny_math off, the complete argument must be a number; an empty argument is zero only when null_eq_zero is on. Scientific notation is accepted. Real-valued arguments also accept hexadecimal notation, such as 0x10 and -0x1.8p2 (16 and -6).

With tiny_math on, the leading numeric prefix is used, and text without one becomes zero: add(12foo,2) returns 14 and add(foo,2) returns 2. Integer arithmetic takes an integer prefix, so div(1.5,1) returns 1. Each function retains its existing signed, unsigned, decimal or floating-point range; this setting does not remove range checks or give decimal arithmetic support for infinity. Numeric input and output use a decimal point regardless of the server’s language settings.

inc() and dec() adjust a signed integer suffix. Without a suffix, they append 1 or -1 only when null_eq_zero is on; otherwise they report an integer suffix error. Overflow reports #-1 OUT OF RANGE. tiny_math does not change these string-counter rules.

Trim argument order: tiny_trim_fun

With tiny_trim_fun off, trim(text,characters,side) uses Penn argument order. With it on, trim(text,side,characters) uses Tiny argument order. The default characters are spaces and the default side is both; l and r select the left or right side.

Workaround. Use trimpenn(text,characters,side) or trimtiny(text,side,characters) when a package needs a fixed argument order regardless of game configuration.

> think trimpenn(xxhixx,x,l)
hixx

Parenthesis groups: paren_groups

A choice. Off by default; importing a PennMUSH database turns it on.

PennMUSH treats a ( that starts no function call as opening a literal group: its commas are text, and its ) does not close the call around it, so cat(x,(a,b)c) has two arguments.
SharpMUSH, with paren_groups off, treats that ( as plain text. The first unescaped ) closes the call and the commas inside separate arguments, so the same call has three.
Why. Whether a parenthesis groups then depends only on whether it follows a function name, and a literal parenthesis is always written the same way. Imported worlds keep Penn’s behaviour so migrated softcode runs unchanged.
Workaround. Escape literal parentheses inside function arguments with \( \) or %( %).

> think cat(x,(a,b)c)
x (a bc)
> think cat(x,%(a%,b%)c)
x (a,b)c
> think cat(x,\(a\,b\)c)
x (a,b)c

With the option on, the unescaped form groups as PennMUSH’s does:

> think cat(x,(a,b)c)
x (a,b)c