never maintened or used since. So... remove from CVS.

Log free disk space errors to sbbsecho.log (as well as print to stdout).

We don't store (findable) hostnames in the file, so don't search for them.
Don't search file if the host is listed in the spamblock_exempt.cfg file.
If you have a really large spamblock.cfg file, this could result in the
unnecessary reading of potential many megabytes of data for each inbound
SMTP connection. We might want to read this file even less frequently and
store the list in a searchable str_list_t or something.

"Application" is not a registered category.
Add TerminalEmulator and Dialup categories.

  I think that > 1MB post data is supported, but the http_request.post_data
  property won't be created if the length > MAX_POST_LEN. Perhaps would just
  store the post data in a file (uh, it already is?) and expose the filename to
  JS scripts? It'd be a lot cleaner than storing the data in a file and then
  reading (or mem-mapping) the file and then copying the contents into a JS
  property.

Allow the JS http_request.post_data property to contain NULs.

open_post_file() will now open the post file (and return the FILE*) even if
  session->req.post_data is NULL, it just won't try to write to the file if the
  post_data is NULL.

mem-map the large post data files using XPMAP_WRITE (read/write) rather than
  XPMAP_READ (read-only) - without this change, this line in read_post_data()
  would cause an exception:
    session->req.post_data[session->req.post_len]=0;
  Now, we seem to have the potential of an off-by-one here (if the length
  mem-mapped is not post_len + 1), but that isn't happening. <shrug>

Fixed a couple of FILE pointer/descriptor leaks if post_to_file() failed.

Changed name of post data file to SBBS_POST.*.*.data (it's not necessarily
html).

So now, uploads > 1MB work, but questions remain:
- wouldn't PUT be a more appropriate method (than POST) for file uploads?
- how can we support post_data > MAX_POST_LEN (now 4MB) in JS?

(don't pass a NULL pointer to fwrite() which can assert or crash).

sub-board (if there was one) which pointed to the cut-sub.

(or deleted) or a group of sub-boards was deleted, and any of the deleted/cut
sub-boards belonged to a configured QWKnet hub, that QWKnet sub-board would
end-up pointing to the next sub-board in the configuration (possibly in another
message group, possibly not part of the QWKnet in question). Now, if a
QWKnet sub is deleted/cut, remove it (correctly) from any QWKnet hubs using
a common (new) function: remove_sub(). The bug was: the remaining sub-board
pointers were moved *before* the remove-sub pointer was compared against the
list of QWKnet hub subs (and NULLed), so while the effort *was* made to address
this problem, it didn't actually work in practice. Thanks for the bug report.

not defined" exception, just report the load() exception - whatever it was
(e.g. could not open load-file).

smb_freemsgmem():
bounce() is using a strange method of copying a message header in memory
(not using smb_copymsgmem) and when it frees the message, ends up with a
double-free of the 2 new smbmsg_t members (not header fields): text_subtype
and text_charset.

- for PETSCII, this is automatic and you don't lose blink, but you do lose
  colored foreground when enabling a bright background ("reverse video")
- for ANSI, this mode is typically referred to as "iCE colors" and disables
  blinking-text support
- Ctrl-AE (^AE) is the new attribute code to enable bright-background
  (will have no effect on ANSI terminals that are not iCE color enabled)
  "E" is now valid in the ctrl/attr.cfg file and string values for JS
  console.attributes assignments, as well
- Ctrl-AI (^AI) - blink - now does nothing for ANSI/ICE color terminals
  (blinking is not supported in combination with bright-background)
- Using a new/non-standard CGA attribute bit-flag to indicate the selection of
  bright-background colors (BG_BRIGHT, bit 10), separate from BLINK.
  This change required all/most char/uchar attribute representations to be
  converted to int/uint.

New text.dat strings:
- PetTerminalDetected (renamed from PetTermDetected)
- PetTerminalQ
- TerminalAutoDetect
- TerminalColumns
- TerminalRows
- TerminalMonochrome
- TerminalColor
- TerminalIceColor
- IceColorTerminalQ
This also moved the MsgCarbonCopyList definition to the end of the file
for now.

PETSCII reverse-video attribute fix:
When a CR is sent to the terminal, the reverse-video attibute is auto-disabled
so update our "current attribute" (curatr) value to match the remote.

Support new printfile/putmsg mode flag: P_WRAP to force an ungraceful
line-wrap (splitting) to the specified column width. If no column width is
specified (0), then this mode will force an ungraceful wrap before the last
terminal column where some terminals *may* auto-wrap.

JS console.printfile() and printtail() methods now support an optional
"orig_columns" argument, similar to console.putmsg(). Must specify P_WORDWRAP
or P_WRAP for this argument to have any effect.

Much improved terminal-type selection/configuration in the user defaults
menu and abort (^C) at any of the yes/no prompts is now detected/handled much
better (to answers to the prompted questions are not saved to the user
settings).

passesd filename is not found - function reports the symbol was not
defined rather than file not found or something similar.

return an error value (-33, it's the magic number).

Borland Implib Version 3.0.22 Copyright (c) 1991, 2000 Inprise Corporation
and MSVC-generated DLLs with exported functions using _cdecl calling convention
... sometimes or with some verisons of some DLLs, implib won't add the
underscore prefix to the exported function name which caused the build error:
Error: Unresolved external '_total_users' ...

Using coff2omf instead to convert the MSVC-generated import library (sbbs.lib)
to the Borland/OMF flavor works, however. I'd seen this issue before but it
went away when the number of symbols in the source DLL changes. Newer versions
of implib appear to *not* have this issue so when we upgrade to a newer
C++Builder, this work-around will no longer be necessary.

When the command is MLST, fp is NULL, so should not be passed to ftell().

backspace/delete key prompt, set the PETSCII flag in the autoterm *and* the
user's setting. If a CBM/PETSCII user actually got that prompt, that means
they were probably connecting to the wrong TCP port in the first place, so
relying on auto-terminal won't work for them on subsequent connections. We
probably should have a way to manually turn off/on PETSCII support, when
needed. <sigh>

^U, ^P, ^T, ^K, etc. and not 0. This allows a CBM/PETSCII terminal user
to actual hit their Backspace/Delete key (^T) in response to the new new-user
prompt

- turn off the user.misc PETSCII flag (too)
- don't copy the current autoterm flags into user.misc - this should no longer
  be necessary since the correct way to detect current terminal capabiilites
  is via calls to term_supports().

- "lsaquo" and "rsaquo" (angled versions of "lsquo" and "rsquo")
- "zwj" and "znnj" (zero-width joiner / non-joiner)

So replace some strstr() calls with either (new local function) strStartWith_i()
or strcasestr(), depending.
strStartWith_i() return length of the matched word, so no need to sprinkle
about magic numeric constants everywhere. The extra calls to strlen() are worth
the code clarity / reliability, methinks.
TODO: find out if there's a way to calculate the length of string-constants at
compile-time (?).

heavy-handed (performs strdup/malloc's and modifications of the strings),
so a temporary hack is to perform a case-sensitive search (using the standard
strstr() function) first. The results won't exactly match the traditional
strstr() and the performance improvement is only for positive matches
(where the correct case was guessed in the passed 'needle' string arg).
TODO: re-write or copy a good/fast strcasestr() implementation for Win32 builds.

(e.g. with ^C).

in clear_convenience_ptrs() - otherwise they can be double-freed when the
message (smbmsg_t) is copied with smb_copymsgmem().

can't because it's a DOS program and the current node doesn't support it.

(when/what is executed, or at least attempted).

The syntax of Multipart/Report is identical to the Multipart/Mixed content type

subsequent parsing (e.g. fall-back to html) to fail.

to a text/html part of a multi-part MIME encoded message if no text/plain part
exists.
Changed smbmsg_t.charset to smbmsg_t.text_charset.
Added smbmsg_t.text_subtype (e.g. "plain" or "html").
Fixed issue with parsing parsed plain-text when the charset was the last
element of the MIME-part header.
MIMEDecodedPlainTextFmt text.dat line now includes the text sub-type arg