1.8.5p2

Version 1.8.5 patchlevel 2 -- Aug 18, 2012

Hardcode:

render_string has been modified to work with multiple output formatters, so it can display the appropriate color syntax for the client. [GM]
Tweak the networking code to try to reduce output latency and queue size. [SM]

256 Color support. Using 24 bit internally for representation and downscaling to the client's possible output values. Use SOCKSET colorstyle=xterm256 to see 256 color support, with ansi(+colorname/+colorname,...) or ansi(#hexcode/#hexcode) for extended color support, or @set me=xterm256. [GM] [MG]
New SOCKSET socket command for changing socket options. Replaces the @sockset command added in p1 for mortals, and can be used at the login screen. Suggested by Sketch. [MG]
@sockset is now @sockset <desc>[=<opt>,<value>[,...]], to query current options or set options for any descriptor. [MG]
The old @doing command has now been removed, and replaced with the normal standard-attribute command (@doing <object>[=<message>]). [MG]

look now takes an /opaque switch, which hides contents listings. Suggested by Wiggles. [MG]
@break and @assert now take a /queued switch, which causes the new <action list> to be queued instead of run immediately, and an /inline switch which gives default behaviour. For Rhost compatability. [MG]
Two completely separate, but equally stupid, bugs made @nsprompt show nospoof info just like @prompt. [MG]
@open, open(), @dig, dig() accept specific garbage dbrefs to use for the exits they make. [SW]

colors() returns a list of all known color names known by PennMUSH, and info about a particular color. [GM]
valid() has two new categories, for validating color names and color code sequences for ansi(). [MG]
etime() formats elapsed times using the same rules as in WHO. [SW]

The meaning of the COLOR flag has been restored. ANSI flag once again means only highlight is supported, while COLOR means all ANSI 8 colors (8*2 including highlight), flash, invert, etc are supported. [GM]
Charset negotiation didn't work on MSVC, and now forces iso-8859-1 when compiled that way instead. Reported by Daniel Powell. [MG]
@config/set and @config/save now log to the wizard_log. Suggested by Israphel. [MG]
The ICLOC attribute, which was set on newly created players but never used in any way, is no longer set. Suggested by Paige. [MG]
@destroy'ing, @undestroy'ing, and the actual destruction of players is now logged. Suggested by Padraic. [MG]
Translation fixes by Stoko.
round(4.5,0) behaved differently on Penn than it did on MUX, and inconsistantly with rounding to more places on Penn. [SW]
The 'On for' and 'Idle' time fields in WHO are smarter and won't overflow their space. Suggested by Ari and Paige. [SW]
Add a few new test cases. Requested by Sketch. [MG]
It's now possible to clear/reset a prompt with @prompt <player>. [MG]
New Pick_Dbref @power for specifying specific garbage dbrefs to use in @create and friends. [SW]

Reordered some checks in tz.c to compile on more distributions. Problem reported by Zen@M*U*SH. [MG]
Setting an exit Transparent didn't cause its destination's desc to get shown as it should have. [MG]
It became impossible in p1 to give a delimiter to wrap() without also specifying a first line width. wrap(%0,%1,,%2) will now use the normal <width> arg for first line width, the same as wrap(%0,%1) does. Reported by Kitai. [MG]
Free PCRE-allocated memory with pcre_free() instead of free(). Reported by ChaosMageX@M*U*S*H. [DC]
Fix some potential memory leaks in timezone file reading code. [SW]
Off-by-one error in the helpfile index code. [MG]
Improve docmentation of ip_addr config option. [SW]
Fix an edge case in ansi_string_replace that could cause an overflow. Reported by Ton Kyrion@M*U*S*H. [DC]
Fix a memory leak in LOGOUT. [SW]