* Command line editing via editline(3) library.
* Context sensitive command and file completion, including remote files.
Enhancements to auto-fetch feature:
* Support for http:// URLs using the http protocol, including proxy HTTP
support via $htty_proxy if it's defined.
* The connection is kept open between successive files on the same host.
(obviously, this does not count for http requests.)
* Return value of ftp is 0 on no error, or the offset in argv[] of the file
which failed (i.e., argv[x] failed, ftp returns x).
* If the path in an ftp URL or classic format line has a trailing '/',
cd to the path and enter interactive mode. Fixes [bin/3011], albiet
requiring the user to help ftp in determining the operation.
Other changes:
* '-P port' works for normal ftp, and is the default for all classic style
auto-fetch transfers and for ftp URLs that don't specify the port.
(previously it would just work for the first xfer.)
* Some code moved into separated files along logical divisions.
* Editing and completion can be compiled out with -DSMALLFTP.
submitted by Heiko W.Rupp.
closes pr-2448
Also, moved Alan Turing's birthday from calendar.computer to
calendar.birthday, removed duplicate Thoreau entry and cleaned up date
ordering in calendar.birthday
of the old zzhack-access macros, which with the current macro definitions
would cause a cast value to be used as an lvalue, which is a gcc extension,
but _not_ what we want.
and address() macros with new macros of the same names (to get values),
and macros with those names but with "set" prepended and which take a
second argument (to set values). The new macros use type casts to do
type conversion, which is much cleaner than using a union and making
endianness-related hacks to get the right 'short's and 'char's from
the right array indices. (YUCK! again, for good measure.)
* in autofetch mode, don't retry a file if the connection failed - skip it
and move to the next file. [bin/3051, Matt Green]
* only print error messages from SIZE or MDTM if the user used 'size' or
'modt' (respectively). prevents extraneous warnings during normal transfers
when connected to a site which doesn't support these (behaviour prior to
last commit), but still allows error feedback to specific user requests
for this info (which the last commit broke).
* 'account': only accept one optional argument
* if invoked as pftp, default to passive mode on (from FreeBSD)
* remove leading '0 ' in progress bar - looked ugly
* use warn instead of perror
* use strncpy when src isn't known to have safe length
* remglob(): use mkstemp() to prevent symlink games, and don't override
_PATH_TMP, use it as the prefix to the temp file
- fix the variable substitution code in make [PR/2748]
1. change s/a/b/ so that it substitutes the first occurance of the
pattern on each word, not only the first word.
2. add flag '1' to the variable substitution so that the substitutions
get performed only once.
***THIS IS AN INCOMPATIBLE CHANGE!***
Unfortunately there was no way to make things consistent without
modifying the current behavior. Fortunately none of our Makefiles
depended on this.
OLD:
VAR = aa1 aa2 aa3 aa4
S/a/b/ = ba1 aa2 aa3 aa4
S/a/b/g = bb1 bb2 bb3 bb4
NEW:
VAR = aa1 aa2 aa3 aa4
S/a/b/ = ba1 ba2 ba3 ba4
S/a/b/1 = ba1 aa2 aa3 aa4
S/a/b/g = bb1 bb2 bb3 bb4
S/a/b/1g = bb1 aa2 aa3 aa4
- add regexp variable substitution via 'C/foo/bar/' [PR/2752]
- add variable quoting via the ${VAR:Q} modifier. This is useful when running
recursive invocations of make(1):
make VAR=${VAR:Q}
will always work... (This may prove useful in the kernel builds...) [PR/2981]
used as lint1 input. That involves having lint pass the '-' through
to the cpp which preprocesses the lint1 input, and having lint1's
scanner recognize a cpp filename "" as "{standard input}".
convert them to pointers. If they're zero, they're converted (to
NULL pointers) regardless of size. If they're non-zero, they can't
be converted (without a cast). This matches the behavior of other
version of lint, e.g. the lints on Digital UNIX and HP-UX.
* recognize that pointers to identical unnamed and untyped structs,
unions, and enums are, in fact, identical. This is done by tagging
each of unnamed and untyped structure, union and enum with a unique
position of creation, which is used as a unique identifier that
when determine whether or not a pair of structures, unions, or enums
are identical.
unions, and enums are, in fact, identical. This is done by tagging
each of unnamed and untyped structure, union and enum with a unique
position of creation, which is used as a unique identifier that
when determine whether or not a pair of structures, unions, or enums
are identical.
* accept the file name '-' to indicate that standard input is to be
used as lint1 input. That involves having lint pass the '-' through
to the cpp which preprocesses the lint1 input, and having lint1's
scanner recognize a cpp filename "" as "{standard input}".
unions, and enums are, in fact, identical. This is done by tagging
each of unnamed and untyped structure, union and enum with a unique
position of creation, which is used as a unique identifier that
when determine whether or not a pair of structures, unions, or enums
are identical.
variable declaration, in addition to within function & variable
declarations and function bodies. I think this comes close enough
to what gcc does to be considered "correct enough." Certainly, it
fixes the problem for the couple of cases where this is a problem
in our tree.
kit, then hacked on by Matt Thomas <matt@3am-software.com>, then by me.
This runs, but it's in serious need of cleaning and/or a fair bit of
reworking. See the README file for more information, and a list of things
to do.
* implement 'progress bar/meter' (inspired by ncftp). use 'progress' to
toggle on. it will display current file size to 5 digits, automatically
determining suffix (up to 16384 P (petabytes) == 2^64).
* 'ls' now uses NLST (unadorned listing), a la older ftp clients. 'dir'
still does LIST (long listing). idea from John Nemeth <jnemeth@cue.bc.ca>
bug fixes:
* return first line of reply in reply_string[] from getreply(), instead
of last line. This fixes [bin/741] (parsing of SYST), and also means
that SIZE and MDTM messages will be parsed correctly if they're longer
than 1 line.
* parse URL-style auto-ftps that have no filename correctly
(e.g, ftp://host, ftp://host/, ftp://host/dir/). pointed out by
Jaromir Dolecek <dolecek@saruman.ics.muni.cz>
* pass the correct size array in 2nd arg of utimes() when setting the
modification time
friedman@gnu.ai.mit.edu (Noah Friedman) made to his modified ftp:
- implement "lpwd" - local pwd
- implement "preserve" - toggle preserving of file modification
times on retrieved files
- allow for explicit "on" or "off" arg to toggle commands
- "exit" synonym for "quit", "msend" synonym for "mput"
- in confirmation mode, allow 'a' (yes to rest of current command),
and 'p' (turn off prompt mode, as if 'prompt off' was done,
effective immediately)
- "modtime" returns time formatted as localtime, not GMT
Bug fixes:
- check for extraneous args on commands
- cleanup const usage, line formatting
- create 0 length temporary file in remglob() to prevent symlink games
(from OpenBSD)
- check length of filename of ~/.netrc (from OpenBSD)
If a user wishes to change a password on a system running YP, and
the master server is not running rpc.yppasswdd, chpass(1) would fail,
even if the user had a local entry. Fix this by checking for local
entry if master is not running rpc.yppasswd iff we defaulted to using
YP (not invoked with "-y").
XXX Unlike the similar change to passwd(1), this one duplicates some
XXX code (makes an attempt to contact rpc.yppasswdd early). This is
XXX a side-effect of the structure of this program. chpass(1) could
XXX use a re-write.
If a user wishes to change a password on a system running YP, and
the master server is not running rpc.yppasswdd, passwd(1) would fail,
even if the user had a local entry. Fix this by checking for local
entry if master is not running rpc.yppasswd iff we defaulted to using
YP (not invoked as "yppasswd" or with "-y").
Printing only those with the name "intr" is too restrictive, because it
means that devices can only have one interrupt, which makes little
sense for many devices and absolutely no sense for several common busses.
- don't echo 'ACCT' parameter when debugging (a la 'PASS')
- Fix checking of directory access for "/foo", the parent
directory is "/", not "" (from FreeBSD)
- remove trailing whitespace on lines
- add any missing NetBSD tags
- cleanups to man page, includinging sorting options description
feature additions:
- variable sized hash marks (from [bin/683], but done in the hash command
as an optional arg)
- more user-friendly transfer time printing (from FreeBSD, with mods)
- '-p' command line option to jump into PASV mode (closes [bin/2857],
but with an option rather than checking argv[0])
- SIGINFO support for printing xfer stats when sending/receiving requests
- '-P port' for changing the port to connect to (from thorpej@netbsd.org)
- '-a': bypass normal login, and try anonymous login (from OpenBSD
via thorpej)
- autofetch files via url (ftp://...) or "classic" (host:/file)
(from OpenBSD via thorpej)
- 'ftp' synonymous with 'open' (from FreeBSD)
-o (the reverse of this, also the default)
- use vis(3) in vputc() instead of handcrufted function (from OpenBSD).
- move gecos expansion into expandusername() (a la sendmail's buildfname).
A generic version of this last bit in libutil would be useful...
- cleanup the code, fix prototypes, etc.
- don't allow quotes in terminal names (for -s and -S) as they confuse
the shell. whitespace is currently forbidden for the same reason.
as there isn't any orthoganol way of escaping this, forbid it all,
which makes [bin/306] unfixable.
- use str(r)char instead of (r)index
Fix PR/2839: su will not build with Kerberos.
- Also:
-Don't coredump when $TERM is not set.
-Add prototypes, remove local old style declarations of system
functions.
-Recognize shells that contain "csh" as being csh alike.
-Don't build with SKEY unconditionally. Obey bsd.own.mk.
make that fail on some machines. Rather, since the code wants to access
the data both ways, use a structure and treat that structure as an array
of chars where appropriate.
