Whoops.

This should normalize everything, with the possible exception of
"zoom" (the thing other OSs call maximized).  Ideally, we wouldn't
allow window size changes when "zoomed" if we could detect the state,
but it doesn't look like we can.

Fixes sourceforge issue 126 (finally!)

Apparently, macOS "maximized" is the same as "fullscreen" in SDL.

Hopefully this is the only lib like this (maybe the resources?)

strListReadFile() can return NULL upon error

Fix CID 509721

Fix for CID 509720

Make ssh_active only true when ssh_session was set.
Only do ssh cleanup if ssh_active is true.
Initialize channel values to -1 (already done in ssh_connect(), but
doesn't hurt anything).

With ssh_sock initialized to zero, if the connect fails, stdin was
closed, preventing ANSI and curses modes from functioning.

First, we need to parse it before load_settings() so we can deal
with it there.

Next, we need to keep the one from the config file available so it
is edited from the program settings, and not the command line version

Keeps the display cleaner when things work and the shift in byte
is displayed.  Should be harmless in cases where shift in is
interpreted per the standards.

Hopefully I'll be able to keep them all wrangled now.

It's likely time to find a nice format that can be used to generate
HTML and PDFs and such now.

Prevents crashes on modern FreeBSD systems that default to implicit
PROT_MAX in the JS code.
(It's possible that +wxneeded will be needed too.. still experimenting)

The first entry I tried adding was "Example..." which was not a
valid hostname, but was copied into the address and looked silly.

Now only copies into address if it's a valid, usable hostname.

Also, add a convenience macro to check if a connection type uses
the network or not.

New functions in netwrap:
bool isValidHostname(const char *str);
bool isValidAddressString(const char *str);
bool isResolvableHostname(const char *str);

We ust the last one (which calls the other two) for this feature now.

Use byte_estimate_to_str() to display upload/download byte totals (better
support for totals in gibibytes, tebibytes, etc.).

I discovered that sbbs_t::editfile() (exposed as JS console.editfile) just
silently failed if it couldn't write to the destination file
(e.g. permission denied by OS).

iniReadFile() can return NULL upon file read error or malloc error. Let's not
crash if that happens.

Even if the empty, iniReadFile() should return an empty list (not NULL),
NULL indicates a file read (or unlikely malloc failure) - so don't write back
what we read and return false (result) instead.

Part of likely work-around of issue #791: it's better to not increment stats
upon a file read error than to reset them all to zero.

Part of potential fix (or work-around) for issue #791

The file position should always be at the EOF after reading the contents.

This should help detect (log errors) upon potential cause of issue #791.

- use a little easier to read .ini style
- add a NULL argument check to local function write_dstats()
  return false rather than deref the NULL pointer

Previously, the partially-read lines would be returned in an allocated
string list. This is a suspected cause of issue #791: even though the file was
successfully opened exclusively (using sopen... SH_DENYRW), it's possible that
Samba had already allowed another client or local process to open the same
file, but is now a denying read (most likely, the first read). As Deuce
pointed out, the xpdev *nix implementation of sopen() locks the region/record
of the entire file. So before this change, strListReadFile() might fail on
the first read, and a function that uses iniFileRead() to modify the contents
of an ini file, might end up writing back the empty list (with added keys),
thus deleting all the existing content of the file.

Also in this change:
- Eliminated the unnecessary local/wrapped str_list_read_file() function.
- Elminate unecessary null-before-free check
- Fixed potential memory leak upon malloc failure (the potentially-allocated
  list wasn't freed).

Using FAT=1 one the make command-line

Revisiting this script due to usability issues reported by W5jsn (TUCUMCAR).

Dynamically figure out the file path rather than storing another property
('path') in the list objects (which ends up in the .ini file). The absence of
the path property (when the file didn't exist) could result in script-ending
exceptions.

.ini files are supposed to be open for read/write access (even when just
writing), so open the file in "w+t" mode in write_list().

Clear console.aborted flag before menu listings: hitting Ctrl-C in some
point could display just header and prompt of lists with no contents.

This should help identify the function(s) used when the corruption occurs.

iniWriteFile() returns bool, not int.

Removed inapplicable function comment.

Hopefully help to determine cause of issue #791 (when is the apparent
corruption of this file occurring?)

This should address issue #787