I discovered the first case when FileBase.get_path() failed, but didn't
(immediately) throw an exception. Reviewing the other instances of
JS_RepoertError() calls found several that either reported a garbage (e.g.
NULL) string value or returned JS_TRUE.

The design pattern used a little in js_socket.c probably should be used more:

	if (JS_IsExceptionPending(cx))
		return JS_FALSE;
	return JS_TRUE;

... but that's more of a refactor than I had the stomach for right now.

It's Transfer-Encoding:, not Content-Transfer-Encoding:, and each
chunk has a \r\n appended to it.

With the latest change, *val will be \n after the last item, so
this avoids an extra loop through the loops.

1 - If cache-control header had more than one option, an infinite
    loop would occur.
2 - For TLS, if the remote closes the connection after the transfer,
    a "Complete" error would occur.

Together, these should fox ticket 191

Pasting mosaics won't work properly, but it wouldn't anyway, so I'm
not too worried about it.

Also, I can't seem to find anything I can paste separated mosaics
into, but that may just be due to Unicode 16 being so new.

Since we're translating in ch[0], don't use buf[j] for mosaic or
we'll lose the translation we just did.

In Beeb mode, receiving the byte on the left prints the one
on the right:

# -> #
_ -> —
` -> £

In Prestel:
# -> £
_ -> #
` -> —

Keyboard mappings adjusted so what you press is what you see.

Prestel and Beeb emulations are different enough that using a shared
switch statement doesn't really make sense.

This adds Mode 7 support for VDU commands 7 (beep), 23 (cursor only),
and 31 (move to X/Y, 0-based).

The rest of the unsupported commands don't seem overly useful in a
BBS context (disable/enable output) or I don't want to deal with
documenting and implementing ("Page mode").

Address warnings from Inno Setup 6, indluding support for both
"administrative" and "non administrative" install modes (!).

This means, if you have not installed SyncTERM previously, you should have an
option to install for "All users" (requiring administrative rights) or
"Just You".

When installing in non-administrative mode ("Just You"), the program will be
installed in your user "Programs" folder, e.g.
 C:\Users\<yourname>\AppData\Local\Programs\SyncTERM

If you previously installed SyncTERM in the common "Program Files" or
"Program File (x86)" directories (which assuming administrative mode, the
only install mode previously), then the installer will install again to
that same location (and require administrative access).

I'm still including a syncterm.lst download option (using curl.exe) because
the Web List feature is not working so great in v1.5. That option should
go away in the next release.

Seen disconnected users "stuck" in irc.js for long periods of time. Hopefully
this helps.

I saw this in my error.log:

evnt DELFILES !JavaScript  /sbbs/exec/delfiles.js line 100:
     Error: -110 loading file 'somefile.name'

Which would terminate the script. We don't need to do that, just log it
and continue on.

Other changes of no consequence.

Specifically, \x11 is not cursor on, and \x14 is not cursor off.

If there's a problem, we'll catch it on the rename(), no need to
add a race condition by checking if it exists first.

Should fix issue 190

There's very little evidence that Prestel terminals supported this,
and zero evidence that BBC Micro terminals did.

ESC and ENQ are ignored like all other control characters in Mode 7
now.

For 10 years (commit e212e2c5), sbbs_t/console.getstr() has limited length
of string input to the available columns of the terminal.

To fix issue #869, rather than change getstr()'s default behavior, add a new
mode flag: K_LINEWRAP which does not limit the string length input based on
the terminal width (and the current column) (e.g. for use with ;string
commands from the default command shell).

Ideally, I'd like to have a marquee-style option (K_mode flag) where longer
strings just side-scroll to accommodate strings longer than the terminal
width, but in the mean-time, this'll do.

So anywhere we think a narrow (e.g. 40 column) terminal is being excessively
restricted in string input width and starting the input in the first column
is not an option/solution, adding the K_LINEWRAP flag to the getstr() call is
the proposed solution.

Fix issue #868 - pretty this string for 40 col terminals

In v1.5, it was changed to sending an ASCII CR instead of an ATASCII
one, breaking auto-login.

No need to use LoadLibrary() and GetProcAddress() here

No need to use LoadLibrary and GetProcessAddress here any more either

Vista ended support before XP did, and it doesn't seem to show up
on any usage lists (while XP still has 0.27% of all Windows systems...
which are 25.46% of systems on the internet)

https://gs.statcounter.com/os-market-share

9 years ago in commit dbbfabf1

Insure that msgcnt is initialized.