Optimization:
No need to read and re-write headers that already have the DELETE attribute.

Fix:
Forced permanent removal of deleted messages (e.g. when creating a new user
that's reusing an existing user number), didn't work unless SCFG->Message
Options->Purge Delete E-mail was set to "Immediately" (not the default).

slyedcfg.js bug fix: Now successfully saves settings if SlyEdit.cfg is nonexistent (not in sbbs/ctrl or sbbs/mods)

See merge request !490

slyedcfg.js bug fix: Now successfully saves settings if SlyEdit.cfg is nonexistent (not in sbbs/ctrl or sbbs/mods)

We don't want our assertions asserting.

Previous commit capped it too early.

The lock is recursive.

Add/use new print mode flag: P_ATCODES for use with trusted format params

Renamed text.dat strings (to better support translation/localization):
- Deleted -> DeletedUser
- Inactive -> InactiveUser

New text.dat strings:
- Deleting
- DoneDeleting
- Deleted
- DeletedNumberItems
- E_Mail
- E_Mails

Maybe Borland is also using pre-vista headers?

Borland hates WINVER

It's basically CLOCK_REALTIME with tv_nsec set to zero.

Like xp_timer64(), except it gets the worst clock, not the best one.
Mostly useful for places where you're using deltas between time(NULL)
to approximate a time, not for if you need to convert to a date or
if you need any kind of precision.

Change internal code of "xtrn-setup" from XSETUP to XTRN-SETUP (we have plenty
of chars available for internal codes now).

As reported by Accession in #synchronet at irc.synchro.net

If this menu file exists, it'll be used to display file information (e.g.
when using 'E'xtended info or 'D'ownload commands) instead of the Fi* text.dat
strings.

Allowing more customization options for file information displays.

datestr() will return this string when passed a time_t value of 0.

FILE_DESC didn't handle blank/no file description correctly

The @-codes that need file size or date needed to call the proper filedat
functions to get those values or else risk them being 0.

I received 6853 duplicate FastCGI error emails today (sent in a 2 minute
time span). I wish I already had this in place. :-(

Caught by Coverity: CID 529977:  Incorrect expression  (SIZEOF_MISMATCH)

Converting the path to UPPER case and then comparing against a lowercase
string would never work.

Bug introduced in commit 6edc0270, reported by xbit (XBITBBS)

Transposed some digits, setting 1056×823 as the resolution instead
of 1056×832.  This resulted on crashes with some output modes, and
"just" the status bar being truncated on other output modes.

Fixes at least part of ticket 178.

This was one of the two major perforance issues with previous
versions of SyncTERM.  Using a larger scrollback will not longer
slow down SyncTERM.

The other major issue is updating the pixel data for every change,
but we can't fix that as long as graphics and text are drawn on the
same plane.

New:
- FILE_COST, file's credit value or "FREE" when applicable
  (see also FILE_CREDITS - the file's credit value or 0 when free download)
- FILE_AUTHOR - file's author (e.g. from SAUCE record) or blank if N/A
- FILE_GROUP - file's author group (e.g. from SAUCE record) or blank if N/A
- FILE_BYTES - file's size in bytes (previously, would use FILE_SIZE for this)
- FILE_CRC32 - 8 hex digits or blank if N/A
- FILE_MD5 - 32 hex digits or blank if N/A
- FILE_SHA1 - 40 hex digits or blank if N/A
- FILE_TIME_TO_DL - estimated time ("HH:MM:SS") to download file at current CPS

Changed:
FILE_SIZE is now the file size estimated in KB, MB, GB, etc., not the exact
file size in bytes.

from text.dat.

Also change the alignment of the Fi* (File Information) text.dat strings to
allow more space for the data fields, allowing better support for adding URLs
(for sysops that want to do so, e.g. by setting FiFilename in text.ini).

At -O3, the only reason the frame pointer will be retained is when
removing it makes it impossible to get any reasonable information
out of a core dump.

I personally prefer to take the small performance hit on affected
platforms and always include the frame pointer because it makes
profiling release builds possible... but I can manually make that
adjustment when I want to.