Changes   [plain text]

Ver .92 -
	Finally figured out how to generate an EOF on a terminal (stty 0 but 
you knew that already ;). Whole module needs a spring cleaning but seems to be
pretty functional. After the (later) introduction of a POSIX stty function the
code involving stty should get a little cleaner.
	Interact, interconnect and expect should all benefit from the EOF fix.
interconnect also looks for FH exceptions, which it treats as EOFs.
Ver .94 -
	Think I fixed the problem of opening /dev/tty. Should work now. Also 
fixed a typo which caused the initial value of Log_Stdout to be set wrong.
Ver .95 -
	Linux had this bizarre problem of stty occasionally not returning when
setting raw -echo. My solution was to run it twice, once for raw and once for
echo. The real solution is of course to write an stty function using the POSIX
module.. any volunteers? :-)
ver .96 -
	Fixed a couple of idiot mistakes concerning DEBUG mode and printing
debug info to STDERR. Changed some debugging spots so debugging info will 
always be printed in readable escaped format rather than raw input.
ver .97 -
	Forced baud rate to get set at startup to make sure it wasn't set to 0
and sending EOFs. Linux now hangs consistently. It will probably not work until
I get to the stty module. Sigh. Time to stop being lazy :)
	Also forced complete handle flushing after handles are opened. Probably
a bit anal but I'd rather err on the side of safety.
	After the stty problem gets fixed I'll probably change over eveything
to use the IO::Pty module. 
ver .972 -
	Fixed setpgrp to be setsid. Oops. Processes opening /dev/tty should be
much happier now.
ver .98 -
	Changed everything. Now requires IO::Tty and uses IO::Stty. This
should make it work on any posix-supporting platform that includes a method
of obtaining a pty. Got rid of the 'detach' stuff, as it cluttered stuff up
too much. Do your own forking.
	Default settings are now done by directly setting Expect::Values, such
as $Expect::Debug and $Expect::Log_Stdout. Things now use 'spawn ids' and
'handle ids' instead of handle numbers. This makes debugging a little more 
	Though there may be more bugs in the ver. I'm hopeful it should prove
to be much more stable, reliable, intuitive and portable than previously.
ver .99 -
	Changed expect() to return the index of the matched pattern + 1. This
enables the 'quick and dirty'   $process->expect($timeout,'patern') || die;
sort of behavior.
ver 1.00 -
	exp_close no longer sends an exp_kill() to the process. This was kind
of a dumb thing to do in the first place. Processes should go away after they
are close()d. In fact, there really isn't any point in using exp_close. Just
use $process->close();. -This isn't true as of 1.03, where soft and hard close
were added to help deal with buffering issues.
ver 1.01 -
	Dropped process still alive during expect. Sometimes the process dies
before the handle finishes getting read. Fixed a typo that caused a response
of 'child process died' if a successfull pattern had a null value.
	Added the ability to do expect() literals instead of regular
expressions. see $process->use_regexps and the package global 
	Fixed internals to work with the documented fashion for using file
ojects. everything is internally represented as ${*$process}{exp_variable} 
instead of ${*$process}{variable}. Pids should be checked with 
$process->pid() now, since $process->{Pid} no longer exists.
	exp_kill() is obsolete. Use kill($signal,$process->pid()).
ver 1.02 -
	Killed the $Expect::Use_Regexps stuff. Changed expect to look for '-re'
strings indicating the subsequent pattern is to be matched as a regular
ver 1.03 -
	Reworked expect() to make the code more fluid. Removed ~50 lines of
cruft. Added soft_close() and hard_close() (see Expect.pod for details).
ver 1.04 -
	Made all handles exp_inited autoflush. IO::Pty does the spawned 
processes for us. All regexp patterns passed to expect() are now 
multiline matched-- this makes matching ^ work for the beginning of lines.
Unfortunately due to limitations in perl matching $ as the end of a line
doesn't work if you are being returned \r\n instead of just \n. In this case
you can use \r?$ to match the end of a line.
	exp_stty now checks to make sure the FH is a tty.
ver 1.05 -
	Added debug level '3'. Fixed/added to tutorial, fixed a couple of
minor bugs.
ver 1.06 -
	Added exp_before(), exp_after(), exp_match(), exp_match_number(),
ver 1.07 -
	Changed the ver by .01 so I could get it up on CPAN :P

ver 1.08 -
	Added multi-match functionality (see docu)  --Roland

ver 1.09 -
	Various small bugfixes: exp_before didn't get set on timeout,
the expect call didn't return on matching EOF, spawn didn't die when
exec failed, Expect crashed in certain cases.

	I cannot believe it: a syntax error in slipped through 
(probably when I changed the version number after testing the patches),
forcing me to do anouther release. >:-(

Expect v1.11

	Bugs fixed:

! max_accum (match_max) finally really restricts the match-buffer-size.

! expect() didn't return the correct error state upon EOF

! soft_close() was re-reading the filehandle even when an EOF
  had already been detected, resulting in a very long delay.

! exp_continue() returned a value that got truncated with certain
  perl versions (5.6 :-( ) on certain systems.

	Features added:

+ Lee Eakin <>contributed a perl version of the
  kibitz script which might be of common interest.  See the
  examples/kibitz subdir.  Thanks Lee!

+ Expect got it's own print() so we can show what's getting
  sent to the spawned program when exp_internal is set.
  (suggested by

+ a session can now be logged to a file by setting log_file().  Thanks to
  Marcel Widjaja <> for suggesting this.

+ I added some aliases for certain methods to help Tcl/Expect users
  to a WLIE experience (Work Like I Expect).  'exp_pid', 'match_max',
  'log_file', 'log_user', even 'send' (which is an alias for 'print')
  are all there now.  But I won't make 'send_user' an alias for
  'print STDOUT', you have to draw a line somewhere.

+ some diagnosis messages now use cluck to print a stacktrace
  (suggested by

+ there is a new option 'restart_timeout_upon_receive', that, when
  set to 1, will restart the timeout within the expect call.  This is
  useful when supervising an application that produces periodic, but
  not well-defined output and you still want to react to certain
  patterns.  Just say

                 [ timeout => \&report_timeout ],
		 [ qr/pattern/ => \&handle_pattern ]);

  (suggested by

Expect v1.12

! exp_Max_Accum didn't work for interact.

- removed soft_close() from DESTROY.  Being overly nice to a doomed process
  doesn't pay off.  Old behaviour is available via
  $Expect::Do_Soft_Close = 1;

! cleanup of log and exp_internal output

+ added various aliases for functions starting with 'exp_'

! moved FAQ and intro into the main pod as I got the impression
  that many users didn't bother to read all the documentation
  or didn't know that it was there.

+ added a hook for log_file: can be set to a code ref.

Expect v1.13

! changed tests to check out pty behaviour (max. string length)

+ added various FAQ entries

+ added autoflush(1) to log_file

+ split 'new' and 'spawn' to be able to set slave pty params via stty
  before actually spawning the program

+ added slave_pty()

+ added print_log_file(), send() now no longer prints to log file or

+ added alarm to to avoid blocking on cygwin.

! spawn() now uses IO::Pty spawn, thus exec errors are reported and
  ssh should work too!

Expect v1.13_06

! spawn is back again

+ rearranged and changed tests to better suit the various systems

+ added rudimentary 'notransfer' option; global only, not on
  per-pattern-basis; workaround available in FAQ

+ timeout handlers now also can exp_continue

+ added 'raw_pty' option, also setting master to raw if isatty()

Expect v1.13_07

+ added and corrected test for exit status; got rid of

! use 'set_raw' instead of stty("raw"); IO::Stty now optional

+ updated docs & FAQs; explained how terminal sizes and SIGWINCH
  should be propagated

Expect v1.13_08

! fixed bug in log_file, parameter now gets set to undef.

Expect v1.13_10 = v1.14

! fixed select in interconnect, may return -1 if interrupted by signal.

Expect v1.15

! fixed bug in _multi_expect, pattern weren't tried against accum due
  to exp_New_Data not set.

Expect v1.16

! fixed hangup with pipe sync upon spawn by adding close-on-exec to pipe handle

! fixed log_file(undef) when logging to CODE ref

! fixed $? mangling in DESTROY by saving & restoring status

! fixed hangup in send_slow

! fixed ugly solaris hack by disabling it for raw ptys

+ added param check for expect()

Expect v1.17

! fixed param check for expect() to allow expect(undef) et al

Expect v1.18

! added another pipe to synchronize spawning. Closing the slave in the
  parent can lead to a hang if the child already wrote something into

! fixed REs in test to deal with shell prompts

Expect v1.19

! fixed non-localized usage of $_

+ added new example

Expect v1.20

+ added early return to send and send_slow if filehandle was closed

+ added test for send_slow

! fixed bug in expect() param handling (exact pattern "0" was ignored)

! fixed bug in _make_readable()

+ now included example in MANIFEST so it gets packaged :-(