Added BL_HELP definition.

compile warnings).

(eliminates bcc32 compile warnings about 'code has no effect').
Also now have a PTHREAD_MUTEX_INITIALIZER defined for all supported
implementations (Win32 mutex, Win32 critical section, and OS/2 mutex).
Now have pthread_mutex_trylock() for OS/2.

which caused lock-ups in RingBufWrite().

readable format.

socket is set to INVALID_SOCKET by another thread.

if the file_name_index or cgi_ext string lists are detected as NULL at the end
of an http session.

string lists are detected as NULL - this is *still* occurring and I'm hoping
this log entry will help determine the cause.

passed for default value (same behavior as iniReadStringList()).

same in a call to sprintf(), likely cause of recent IndexFileNames weirdness
in web server.

fix the segfault on "save bbs" thing.  If it does, there's still a problem.

highwater/timeout being innefective when the ringbuffer gets many small
writes very quickly (ie: During the "Synch" logo display).  For some reason,
this seems to happen most on fast connections (Possibly the socket
implementation manages to combine the write()s on a slow connection)

Use ftpalias.ini instead of ftp_alias.ini.

(optionally) gets its length, all in one function call.
crc16(), crc32(), and chksum()
Some global methods that would silently fail if JS_GetStringBytes() failed,
will now generate an error.

number generator.

object/scope hierarchy.