Compatibility Arguments

Functions that accept a different number of arguments here. Every one of these is additive: the call you would write for PennMUSH keeps its PennMUSH meaning, and the extra argument is optional.

SharpMUSH.Tests/PennMUSH/function-arity.tsv holds PennMUSH’s own declared arity for every function, and a test compares it against the registered arity on every run, so this list cannot silently grow. See docs/guides/registry-parity-and-coverage.md.

A trailing `<precision>` on the time functions

A choice.

Affects convtime() convutctime() csecs() etime() etimefmt() idle() msecs() secs() stringsecs() timestring() uptime().
PennMUSH answers whole seconds.
SharpMUSH answers whole seconds too, unless you ask for another unit in one more trailing argument: s (the default), ms, f for fractional.
Why. SharpMUSH stores creation and modification times to the millisecond, so csecs() truncating to seconds would make [num(%0)]:[csecs(%0)] fail to reconstruct an objid. Rather than change what a PennMUSH call returns, the finer value is a separate request.
Workaround. Omit the argument for PennMUSH behaviour.

> think secs()
1789000000
> think secs(ms)
1789000000123

(The numbers are whatever the clock says; what matters is that the second is the first with three more digits.)

A trailing `<osep>` on the vector functions

A choice.

Affects vadd() vsub() vmul() vdot() vmax() vmin() vcross().
PennMUSH joins the answer with the same <delimiter> it split the inputs on.
SharpMUSH takes a fourth argument that overrides it, defaulting to <delimiter>.
Why. Every other list function in the game spells the same idea that way.
Workaround. Omit it.

> think vadd(1|2|3,4|5|6,|)
5|7|9
> think vadd(1|2|3,4|5|6,|,%b)
5 7 9

Prepared-statement parameters on `sql()` and `mapsql()`

A choice.

PennMUSH takes at most four arguments and rejects a fifth.
SharpMUSH binds every argument past the fourth as a prepared-statement parameter, substituted for a ? in the query.
Why. The alternative is softcode concatenating values into query text.
Workaround. None needed; a four-argument call behaves as PennMUSH’s does.

> think sql(lit(SELECT name FROM people WHERE id = ?),%b,%b,%b,7)

A trailing `<count controls>` on `strlen()`

A choice.

PennMUSH takes one argument. A second one is joined to the first, commas and all, with a warning that this is deprecated: strlen(a%tb,0) warns, then answers 5.
SharpMUSH takes an optional second argument. A false value counts control characters such as %t and %r as zero, as displaywidth() does, so strlen(a%tb,0) is 2. Omitted or true, each control character counts as one, as in PennMUSH. (#1256)
Why. A tab or a newline takes up no columns, and code that pads a table needs that width.
Workaround. Omit the argument for PennMUSH behaviour. A PennMUSH-era call that relied on an unescaped comma being kept needs it escaped: strlen(a\,0).

> think strlen(a%tb)
3
> think strlen(a%tb,0)
2
> think strlen(a\,0)
3

For Latin-1 text the two servers count alike. Beyond it, PennMUSH counts the bytes of the UTF-8 encoding and SharpMUSH counts display columns: a wide CJK character is two and a combining mark is none. [chr(28450)][chr(23383)] is two wide CJK characters (the word “kanji”); PennMUSH would answer 6 for those two characters:

> think strlen([chr(28450)][chr(23383)])
4

Argument-less forms PennMUSH does not offer

A choice.

PennMUSH answers stext() with #-1 ARGUMENT MUST BE INTEGER, and declares lwhoid() with at most one argument, so lwhoid(me,online) is an argument-count error.
SharpMUSH treats stext() as stext(0), the innermost switch, and lets lwhoid() take the <status> argument lwho() takes.
Why. An omitted level means the innermost one everywhere else. PennMUSH serves lwho() and lwhoid() with one C function and only declares them differently.
Workaround. Write stext(0), and give lwhoid() at most one argument, for code that has to run on both.

> think switch(a,a,stext())
a
> think switch(a,a,stext(0))
a

`ansi()` and `lit()`

PennMUSH spells “the last argument swallows the rest, commas and all” as a negative maximum in its function table. SharpMUSH reaches the same result with a literal flag on the function instead, so the declared maximum is not a comparable number. Behaviour matches; only the declaration differs.

Compatibility Arguments