- fix bug in wcs2any() that threw away all the string after an invalid/unconvertible sequence instead of just the invalid/unconvertible characters
- fix test main() to use new function signatures

New abstractions for writing to the pipes, and changes to use wchar_t in many places. First attempt to make stuff write to the pipes in UTF-8.

The idea is to replace all the IPC code with invocations of this abstraction, and then replace the IPC with something saner, like sockets, or even the Marvin protocol.

A hack to make messages work properly for now. A pid_t was being put in an int which was assumed to be 4 bytes long, then memcpy'd into a string. This patch forces it to be 4 bytes long by using a uint32_t.

This might mean lossage with a 64 bit pid_t, in the very, very unlikely scenario that the pid is bigger than can be held in 32 bits. A proper fix would be to get enough x's to hold a pid_t however long that happens to be, but that would be Effort :)

There is now an "ison" function, please use that instead of grovelling through the who file in future.

updated for 2.14.2

- Changed %d and %u, where they referred to size_t things, to %zu
- Changed unsigned int to unsigned long int where it was the target type of a cast from a pointer

These fix warnings when compiling on x86-64.

mud rooms from /usr/local/lib/mw that never made it into cvs before

fixed dumb bug where I tried to compile the wrong file

- Moved the gag tables out of bork.h into a source file, so the
executable now only has one copy. They're now const.
- Now using gnu99, which people seem to have been coding to anyway,
with GNU libc extensions
- removed #define from iconv.c which the above made redundant
- Optimisation flag is now in CFLAGS instead of DEFS (makes more
sense there)

Next step, check if it compiles on AMD64.

add spaces

*am am am*

fixed a bug in the jive gag where e.g. "I've" -> "AH''ve"
(now "I've" -> "ah've")

fixed mismatched parens in credits

updated for new release

expected, now catch that and start from the beginning instead

oops, put "and" in the right place this time

Added myself to credits()

oops, forgot changelog

Fixed permissions for the colour directory (though the !colour
command shouldn't really have used access() in the first place)

updated for 2.14.1

bumped up tweak version number

- Fixed makefile bug stopping command.c being compiled when it changed
- Added commands to change locale and character set, so bbs
users can get sane terms

RPMs coming soon

init_locale() now defaults to C / ASCII locale if locale env
variables are absent or empty

removed utf-8 hack
added sanity checks to charset changing code
added function for querying current charset
added locale init function
called locale init function from main()

The plan now is to add a command that bbs users can use to change
their character set to whatever their term is using.

New source file containing functions to convert strings between the
local charset and wchar_t, and one to change what "local charset"
means.

oops, corrected typo

Rewrote strip_str() in a rather more locale-flexible way

Workaround bbs user locale issues (set LANG to en_GB.UTF-8 by default)

Bumped up version number to 2.14.0

added "0;" to non-bold colour sequences so they work as expected

Changes in the latest rpm, and doc changes to match.

COPYING, INSTALL, LICENSE and README now get installed to the mw dir.
Also changed the specfile to reflect this.

Some documentation for packages:

LICENSE is the GNU GPL v.2 verbatim
README is copyright and credits
COPYING is a brief description of the license (largely just the form at the end of the GPL)
INSTALL is instructions on building and installing, including creating a user for mw to run setuid as and anonymous access via the bbs user.