New bug, reported by Melkor (STARFRON).

The host portion contains ":" after split_port_part(), resulting in a 400
error.
Renaming is_legal_hostname() to is_legal_host(), since requests to webservers are to
"hosts" not "hostnames".

MSVC Error D8016. Just disable "edit and continue" support.

Also added a .gitignore file.

Have msgtoqwk() return a negative number upon error (e.g. blank message, failed to read to body text), which means *no* message header was written to the messages.dat file, so don't increment the record (block) counter used to create the index (NDX) file.

If msgtoqwk() returns 0, that means that a header was written, but no body text (e.g. a vote message) and that is different than an error.

When both Indexes (NDX files) and vote messages are included in a QWK packet (an unexpected, but valid combination), the "Record number" stored in NDX files for subs with votes would be incorrect as it did not increment when a vote message (with no body text) was added to the messages.dat file.

Reported by user that was trying to use MultiMail to read a QWK packet that included both votes and indexes and "IgnoreNDX: No" in their .mmailrc file.

putmsg(), used to display messages and display/menu files, will no longer pass-through the ANSI sequence ESC[c (found in some corrupted ANSI posts to fsxNet->FSX_BOT echo) to the client terminal since this will stimulate a (unwanted) response from the client.

I do wonder if the outchar_esc value check should be == here instead of >=, but I'm not familiar with the string and SOS states. I wonder if any display files sent from the BBS would ever include those sequences.

No functional change with regard to the ANSI output state, just making the code easier to read.

Removed the conversion of `[ and \xFA[ to \x1b[ in putmsg(). This unexplained output translation has been supported since at least v2.xx and I have no recollection of exactly why it was added. If I recall correctly, some BBS software at some point in time sent ANSI-encoded messages into message networks by translating the ESC (\x1b) character in the ANSI sequences to either ` or \xFA in the process. This "feature" seems like it would still work, but it's completely undocumented and I have no idea why it was added (and seriously doubt anyone relies on this behavior), so I'm removing it. If anyone misses this feature, I'll happily add it back, but I'm skeptical that'll happen.

When reading messages, circumvent ANSI-encoded messages by replacing the ESC
character (ASCII 27) with '.'. This could be made an optional behavior if
someone really wants ANSI-encoded messages to be read via smbutil.

Also, allow continuous reading of 1 or more consecutive messages using the
-<digit> option (no prompting).

<nelgin> However, dm should probably maintain it at the source.
<nelgin> DigitalMan, I've uploaded gtkuseredit_glade.c which also widens "Sex" to 5 chars
<nelgin> I've also uploaded my .glade file for comparison
<nelgin> and I've uploaded my convert perl script.
<DigitalMan> the sex/gender field is a single character
<nelgin> Yes, but it doesn't display properly unless I widen it to 5.
<DigitalMan> ah

The password was (recently) limited to 8 chars, so that made sense.
The 24 character limit of the real name does not make sense and I can only
assume it was a typo.

- DATE: (alias: TIME:)
- BDATE:
- PWDATE:
- SINCE:
- LASTON:
- FIRSTON:
- LASTNEW:
- EXPDATE:

Truncating. str is a pointer, so sizeof(str) is only 4 (or 8) bytes.

e.g. MM/DD/YY for drop files that assume that format, or either MM/DD/YY or
DD/MM/YY (depending on sysop preference) for backwards compatibility in
expanded @-codes, Guru chat keyword.

Fixed the birthdate check while logging on to work with new/all birthdate
formats.

... and up to 200 years old (hey, I'm optimistic!)

Also, fixed get_date(), it returns 0 on success, but the success return logic was inverted:
it only returned 0 if the date was unchanged from the previous value.

Note:
It would be nice if this user editor prompted to save changes if changes have been made a new
user is selected (e.g. with the arrow buttons).

These allow the easy reading or writing of these sub-field values of the
user.birthdate property. When migrating from the legacy formats (e.g. MM/DD/YY
or DD/MM/YY), it's required to write all 3 properties to get a correct
birthdate/age. Otherwise, "13/31/69" could become "19691/69" (for example)
which isn't going to parse correctly.

'i' is the return value of savemg(), used to determine success or failure
(and what kind of failure) - so we can't use that for the usernumber of
the recipient.

Support the old MM/DD/YY or DD/MM/YY format for reading. New users get the new format.

strcpy -> SAFECOPY
sprintf -> safe_snprintf or SAFEPRINTF2
strcat -> SAFECAT

Enabling this feature causes the node's temp directory to be cleared before
running the external program and causes the drop file(s) to be created-in and
read back from (when appropriate) the node's temp directory.

Creating drop files in a node's temp directory eliminates the possiblity of
an external program (e.g. door game) accidentally reading the wrong drop
file (e.g. wrong-cased name or a stale file of a different type).

Note: Synchronet/XSDK doors *always* read their drop file (xtrn.dat) from the
node directory, so don't try to use this option with those doors. Only doors
where you can configure the drop file path or specify on the command-line
(e.g. %g for the temp dir or %f for the fullpath of the drop file) should
potentially use this new feature.

While SBBSecho would notify local users of imported messages addressed to
them (both EchoMail and NetMail) and QWK-imported messages recently gained
the ability to notify local recipients of messages addressed to them, other
methods of posting messages (e.g. just using the normal terminal server
method, e.g. the 'P' command) or using any JavaScript (e.g. web) methods of
posting messages lacked recipient notifications. So that's now fixed.
And for JavaScript-saved local email messages too. I don't *think* this will
result in any duplicate notifications of received messages, but let me know
if you see excess/duplicate notifications.
Anonymously posted messages are excluded.
Messages posted to yourself are excluded (who does that?).

e.g. FILL:\x20 to fill with space (for Michael Long via USENET)

NETMAIL: user's personal netmail or email address - for WitNik
FWD: "On" if the user has opted for forwarding their mail to netmail or "Off"
TMP: user's chosen temporay file type (e.g. "ZIP")
NOTE: user's note (same as MEMO1)
LASTIP: user's most recent IP address at time of logoff

The return value of getlastmsg() is displayed as the total messages in a
sub-board when scanning subs. For sub-boards that allow voting, use the new
smb_msg_count() function to determine the number of displayable messages.

If you have a stored errno value (or Socket.error), these are how you convert
that number to an error description.

Socket error descriptions on Windows can be quite long (e.g. > 128 chars) and
include punctuation, so include at the end of logged socket error messages.

If FormatMessage() fails on Windows (in socket_strerror()) return the
GetLastError() value in the error description.

Socket error descriptions on Windows can be quite long (e.g. > 128 chars) and
include punctuation, so include at the end of logged socket error messages.

Vote messages and poll closures are not displayed to users as messages, so
exclude them from the total "posts" returned from getposts(). This uses
the new smb_msg_count() function which is slower than simply checking the
size of the index file (*.sid) and dividing by the index record size. So
for sub-boards that don't support voting, use the old/faster scheme.

The inferred message type should match the stored message type. Confirm that.

Don't exclude polls form the message count used for max_msgs enforcement
during maintenance.