output of nodestatus() (e.g. used by Synchronet Control Panel node window).

directing the log output accordingly (apparently online is not always being set
to ON_LOCAL as expected, checking node_num is more a consistent method).

Don't prompt for "delete mail from everyone" when there is only one e-mail
waiting for the user.
Use some existing text.dat entries (some old and some recently added) where
identical static strings were still being used for no apparent reason.
Added several text.dat entries for the QWK settings menu, node status, and
some other single-line messages (replacing static strings).

be replied to (e.g. with a bounce message) - used by the list server.

Type mismatch in parameter 'optlen' (wanted 'int *', got 'unsigned int') in function nonblocking_connect

comparing in matchuser() - so "digit alman" matches "digital man", etc.

Get the last socket error instead and return that (not tested on Win32)

SENDER field with trailing whitespace removed.
Added hash file validation (checks message numbers, times, source type values).

Ragnarok - sub_list would be populated for sub-boards current user did not have
access to (grp_list element did not exist, but subs could still be accessed via
msg_area.grp) and can_read property did not check group access.
Created userdat convenience functions for checking access/read/op rights to
sub-boards.

the status bar directly (this is now handled in the UpTimerTick), thus
preventing crashes if 2 callbacks are called concurrently.

* double spam.log entry ("TAGGED" and "IGNORED") for the same message
* DNSBL_IGNORE option did not work unless the DNSBL_SPAMHASH was enabled

return character rather than a large number of backspace characters.

Change matchuser() to ignore white-space as part of name/alias matching.
Change usermailaddr() to strip white-space from user names that include both
dot and space.

BTW, there is this interesting msvc6 warning:
xp_dl.h(17) : warning C4229: anachronism used : modifiers on data are ignored

threaded project too?

which are a result of the OpenDoors API design.

Mostly they seem to be using unsigned long rather than DWORD for Win32 API
calls.

Need to check other Borland distributions and see if BDE shipped this way as
well.

than dynamically allocated string list. Also, allow a blank JavaScriptLoadPath
to be specified in the sysop's sbbs.ini if they wish. This resolves the memory
leak issue with this startup parameter.