left-justified server prefix), e.g. "term" now added for terminal server
errors.

you can't print and free it again after calling that macro.

and send_mlsx().
Fix numerous printf-style argument bugs, many in the TLS and MLSx support.

Put "TLS" in the failure log messages rather than repeating "FTP".

Windows debug build.

- lower log level (raise severity) of "Waiting for ..." msgs from Debug to Info
- log a message when done waiting (for clients, threads, etc.)
- only one #### log line per termination event

-- Albert Einstein

(the multisock stuff already logs this for each interface and port).

Things return allocated pointers with errors in them now.
FTP server updated, the rest to follow "soon".

Call FULLPATH() on local directories after changing
Add cdir to MLSD in the bbs: mount
Return full path in MLST

(dirs and files).  Include the unique fact in aliases as well.  Clients can
now tell that aliases are the same entry.

For MLST <dir>, always return the full path of the dir instead of the
specified or relative path.

Likely some other minor fixes as well.

like symlinks... note that they aren't shown as symlinks in the directory
output, so it may be a bit of a surprise when relative paths are relative
to the target rather than the alias.

to handle this somehow (but currently doesn't).
Also, remove some debug output.

Add the '..' parent directory to libs and dirs
Fly the TVFS flag now that relative dirs are working

LIST should likely check the return value now...

Add support for OPTS MLST command.
Fix MLST FEAT response.

in the other stuff too.  The local mounts may not be at 100% either, but
the appear to work.

I wasn't expecting to do this, since SFTP is often viewed as "better", but
SFTP doesn't have a specification, the implementations don't follow the
specification drafts that are out there, and the whole sharing a port between
file transfer and terminal server thing is irritating.

This has only been tested with lftp, but it seems to work.

it's non-zero.

HANDLE_PENDING() contains a return, so it's theoretically possible that the
memory allocated by the previous JSVALUE/STRING_TO_... allocation could
be leaked. So now we pass an optional pointer to HANDLE_PENDING() which will
call free() on it if it's not NULL, and then sets it to NULL for good measure.

Lower likelihood that ftp_remove() will log a false error.

accepted (e.g. "EPSV 2"). For now at least, accept any arguments after EPSV
(and treat them all the same).