New in 3.10.5r:
 * for random batch test derivation (torx --batch )
   remember when all incoming transitions were quiescence;
   if so, prefer to stimulate instead of observe
 * changed Trojka to avoid loosing a non-zero random number
   seed on linux (due to the fact that on linux,
   but not elsewhere(?), srand and srandom seem to be
   the same function, and we have two seeds in Trojka,
   one of which is most often 0, and given to srandom,
   after the srand() call has already been done

New in 3.10.5q:
 * spex explorer: bugfix: changed some fixed string array
   sizes to allow for the relatively long strings that are
   needed to represent large (negative) integers

New in 3.10.5p:
 * in torx (driver) ignore predicates that are passed straight
   through from the menu, to allow instantiator to attempt
   to instantiate predicates associated with the event in the menu.
   special-case to improve user-experience when selecting an
   input with predicates in the menu and clicking 'selected input'

New in 3.10.5o:
 * made torx more robust w.r.t. events given to 'input'
 * in spex changed predicate separator from , to ; 
   (like we have elsewhere)
 * minor changes to spec source to avoid compiler warnings
   while compiling a primer
 * added spex-instantiator
 * made adaptsim more robust: for output only consult
   instantiator if predicate list of output action that
   is to be returned is non-empty

New in 3.10.5n:
 * more changes to get working again on cygwin xp
 * fixed nasty anifsm buglet (invoke function with
	wrong number of parameters in a place we do
	not always come, or sometimes we do)

New in 3.10.5m:
 * more changes to get working again on cygwin xp
 * interprocess communication over tcp;
     firewalls may mess that up, or so it seems...
 * TODO: make sure all mcast connections to
    tmcs are broken so it can exit when no longer needed
    (add some debugging to get all present connections out?)

New in 3.10.5l:
 * freeze partial changes to get working again on cygwin xp
 * to be completed in  3.10.5m

New in 3.10.5k:
 * minor fix in xtorx to support callback names containing '-'
 * changed implementation of random function in primer
 * disabled lots of debug prints in primer

New in 3.10.5j:
 * changed torx(1): do not say 'verdict pass hit' when we
    actually only have 'verdict pass'

New in 3.10.5i:
 * changed intersector(1): default value for CHOOSEINPUTS
    is now 1, which is more likely to be what is wanted --
    which means that with this new default it is no
    longer needed to configure it ourselves.

New in 3.10.5h:
 * fix primer to no longer return IOKIND answers that suggest
    that it can do things that it actually can not (because they are
    not in the menu) - earlier attempt at fix in 3.10.5g
    was incomplete
 * xtorx now recognizes 'pass' verdicts also without 'hit'
 * intersector no longer shows funny weird added menu
    entries in the intersection result when we have only
    a single attached thing (primer/guide/test)
 * intersector now distinguishes between empty menu
    because of inconsistency, and empty menu that is
    just empty because all constituents are empty

New in 3.10.5g:
 * fix in torx(1): flatten list of INSTFLAGS
 * added missing dependency in Src/Explorer/spex/Makefile.am
 * fixed tracker: updated to use new em/tm lib API
 * in primer: no longer add instantiation result to the menu,
    because we cannot (currently) remove it from menu,
    so if  instantiation is refused by the adapter it also
    stays there - but then, when we retry after failed
    stimulus it might be chosen again, and again, and again...
 * iokind return value of primer now tells driver to stop when
    primer menu is empty, intended for execution of batch
    (offline) derived test cases

New in 3.10.5f:
 * integrated promela spex updates
 * from primer removed debug print that upset primexp
 * improved anifsm(1) attribute box resize behaviour

New in 3.10.5e:
 * changed autexp(1) connection matrix output. it was a boolean matrix.
    now it contains the number of transitions between states.

New in 3.10.5d:
 * fix in primer: omitted em/tm pattern identifier sets

New in 3.10.5c:
 * updated spex promela extension syntax.
   we now have two main primitives, spexchan and spexio.
   
   spexchan declares a named spex channel that is used
   to communication with the environment.
   
   spexio interacts with the environment.
   it gets a spex channel name and zero or more parameters.
   we distinguish between 'in' parameters and 'out parameters.
   
   'in' parameters consist of the name of the variable that has
   to be updated with the value of the parameter exchanged with
   the environment. optionally, a constraint can be imposed
   on the value that is exchanged.
   an 'in' parameter is denoted by a prefix '&' in the spexio
   parameter list
   
   'out' parameters give the value that is to be communicated.
   the value is given as literal number, or as mtype, or a
   variable is given from which the value has to be taken.
   
   example:
   
   spexchan aap = [0] of { int };
   int v;

   spexio(aap, 5); // communicate 5 on chan aap
   spexio(aap, v); // communicate value of v on chan aap
   spexio(aap, &v); // set v to value exchanged in communication
   
 * extended Common Tcl libraries em and tm with explicit
    pattern set handles to be able to have multiple pattern sets
 * intersector writes errors to standard error instead of standard out
 * intersector passes on INPUT/OUTPUT attributes that it gets
    from its connected primers or guides
 * intersector evenly distributes INPUT/OUTPUT weight attribute
    that is associated with a label over all events that match the
    pattern
 * updated partitioner with command line options to
    allow simulator mode (traces algorithm), and primer interface
    commands GOTO, STATEID, GETINPUT, GETOUTPUT,
    multiline STATS
 * extended primer to allow labels in which attributes are
    embedded (as: "label [weight=10,pat=aap]" )
    and extract the attributes and explicitly give them
    on the primer  interface
 * added new component: tracker.
    it works 'on top of' a primer, and is intended to keep track
    of where we have been in the primer.
    it can be configured to recognize a label that resets to
    the start state, and it tries to avoid going where it has
    been before, and it uses weights to achieve that.

New in 3.10.5b:
 * in the torx(1) keep track of current super-state in primer
    to avoid superflous requests to primer to change super-state
    (via C_GOTO) to the state it already is in
 * for debugging purposes of torx --suite, state information
    is written to file /tmp/states.<PID> where <PID> is the
    process id of the torx process
 * added mkprimer(1) support for the CADP-based LOTOS
    explorer to the distribution (still needs manual page)
 * additional debugging output in primexp(1)
 * changed primexp(1) to better deal with long lines
    it may receive from the primer connected to it
 * additional debug output + minor fix in allinst
 * added caching of explorer results to primer(1)
 * during unfolding the caching code appends new
    transitions to /tmp/primer<PID>.aut
    such that it contains the automaton that has been
    explored on-the-fly so far.
    NOTE: the initial 'des' line is incorrect, however,
    autexp -d /tmp/primer<PID>.aut > /tmp/primer<PID>.dot
    will be able to generate dot output that can be used to
    analyse the graph
 * added additional check to avoid adding superflous state
    to list of states to be expanded to primer(1)
 * additional debug output in primer(1)

New in 3.10.5a:
 * allinst: new version of instantiator that returns all
   possible instantiations
 * other changes in primer for torx --suite
 * extended autexp(1) with a -e flag to include edge 'names'
   in the edge labels (so they show up in a visualization)
 * changed anifsm(1) Open (and Save?) menu commands
   to use the directory containing the dotfile shown as
   the initial one in the file selection box

New in 3.10.4:
 * improved (fixed?) intersector(1) GOTO command
    (used when intersector is used for batch derivation)
 * improved (fixed?) intersector(1) dealing with
    var_type fields in events: var_type$nr now works.
 * bugfix in intersector(1) (debug print referred to
    non-existing variable)
 * added (early version of) spex explorer (no man page yet)

New in 3.10.3:
 * so far Trojka (Promela Primer) would put a limit on
    the number of consecutive internal steps
    (maximal depth) it would explore
    (the default used to be 100);
    now the default is 0 (meaning there is no limit)
 * fixed -m flag for Trojka-generated primers such that
    a different maximal depth can be given
 * minor tweak to torx(1) --suite to output the
    test cases under construction with pass leaf nodes
    when the intended depth is reached

New in 3.10.2:
 * changed (enhanced?) the implementation for
    torx(1) --suite . it now uses files on /tmp to store
    intermediate levels during expansion.
    these files tend to be big.
 * added experimental torx(1) command line option
    --suite-rec  that during expansion detects recursion
    through the root of the test case (start state of spec)
    and then in the 'linear form' labels them with a 'R'
    node type (no support for this in lin2aut yet)
 * changed torx(1) command line option --batch-automaton
    to 'include' the --batch option
 * (kind of) documented --suite and --suite-rec in torx(1)

New in 3.10.1.1:
 * New, much more open License:
   we distribute TorX now under the Apache License, Version 2.0
 * except for the license this version is equivalent to 3.10.1

New in 3.10.1:
 * added --suite to torx(1) to systematically generate
   a (prefix of an) exhaustive test suite on standard output
   in a 'linear form' of a test tree (test tree contains no fail)
   on standard output.
   combine with --suite  with  --depth 3  
   (or some other depth) to give an upperbound on depth.
   implementation (naive anyway) is even more naive
   then already intended (and thus even slower than
   intended. suggestion for improvement in the code)
 * added lin2aut(1) (no man page yet) to translate linear
    test case treee form (read on stdin) to .aut on stdout
    (use: torx --suite batch.if | lin2aut )

New in 3.10.0:
 * implemented simulation (-S sim) for promela primers
 * promela primers no longer print arbitrary messages
   to standard output
   (to avoid breaking the primer-driver interface)

New in 3.9.4:
 * fixed torx(1) --batch to generate correct des line at start
   of generated .aut file

New in 3.9.3:
 * fixed torx(1) such that with --batch we now only have
    the 'real' label in the generated .aut file
    (without the 'channel name', 'is_suspension' and 'preds')
    i.e. this fixes the TODO of 3.9.2

New in 3.9.2:
 * fixed torx(1) such that --batch now does work (again)
   (somehow we broke it?) (TODO: have better labels (omit
   unnecessary stuff) in generated Aldebaran (.aut) files)

New in 3.9.1:
 * made primer(1) slightly more robust against torx-explorer(5)
   programs that write unexpected messages on their standard output
 * bugfix in anifsm: added forgotten routine that print warnings
 * bugfix in mctrlsrv: fixed potential expr error
 * bugfix in anifsm: autexclude the 'name box' that is created
   by log2anifsm, and do not use the label of that box when
   computing the common prefix of the node names, such that
   when we save to .aut the dynamically generated automaton
   of the test run the saved automaton does not contain that box,
   and does contain numeric state names
 * extended log2aut and log2anifsm to be able to function
   even when the log does not contain STATEID lines
 * bugfix in anifsm: catch error message when closing pipe to 'ps'
    (when used to check if anifsmsrv is already running)
 * extended anifsm with commands 'ctl render off' and
    'ctl render on' to disable/re-enable rendering, to be able to
    make several dot layout commands appear 'atomic'

New in 3.9.0:
 * minor formatting (troff) fixes in many man pages
 * in torx-intro(1) reorganised list of commands
 * in torx-intro(1) added piece about LTS
 * in torx-intro(1) added piece about external toolkits
 * added man page mkprimer-jararaca(1)
 * only build TorxOpen subdirectory (with torx_open(1))
   when configure detects the presence of CADP
 * minor fixes in Trojka to compile on Mac OS X (Darwin)
   (among others: now include stdlib.h instead malloc.h)
 * to allow Trojka to allocate more than 4 Mb of memory
   on Mac OS X (Darwin) we now use malloc instead of
   our (old spin's) own sbrk-based allocator
 * fixed xtorx(1) to not loop forever during startup on
   Mac OS X (Darwin)
 * fix in anifsm(1) to work correctly when a 'dot ...' command
   is send for a cloned window. also made the updates happen
   at the same time when the Render button is pressed
   (this was already in 3.8.1 but not documented here)
 * INCOMPATIBLE change in anifsm(1): in the 'dot edge'
   command we now use a single word for the edgelist
   (which now consists of a list of node names separated by ->)
   instead of the two words, one for source and one for dest,
   that we used previously.
   so now we use 'dot edge src->dst label l name e0'
   instead of old 'dot edge src dst label l name e0'
   This is compatible with anidot(1) which is anifsm(1)
   extended with editing capabilities copied from GraphViz'
   doted.
 * updated log2anifsm(1) to generate above syntax
 * updated anifsm(1) with graph editing capabilities of
   anidot. This also allows one to edit a graph and then
   write it out to file in Aldebaran (.aut) format.
 * added 'media controls' to mctrl(1) to automatically
   (re)play an animation
 * increased (expect) buffer size in adaptor(1) to be able to
   deal with 'lots' of output, as for example is 'generated'
   by the confprot iut when coverage data is produced.
   (to avoid 'buffer full' problems we previously also
    encountered with xtorx and with torx)
   (NOTE: the actual buffer size chosen is a wild guess)
 * in torx(1) changed the way we pass the suspension
   event received from adapter to spec. So far, if a
   DELTA label was configured in the .if file we would
   always use that (even when the adapter gave a
   different label). From now on we use the label given
   by the adapter, except when it gives the empty string,
   only then we use the label configured in the .if file.
   (documented in torx-config(4))
 * added man page torx-hostname(1)
 * extended torx-explorer(5) with details about fields used
   in symbolic interface (preds, freevars etc.), and added
   example with 'i' command and more 'e' commands
 * changed autexp(1): extended event number of initial
   state with edge info to be able to highlight initial transition.
 * for jararaca(1) added dependency in Makefile to trigger
   generation of version.h when compiling main.c
 * extended ltsaexp(1) (the ltsa-based explorer for FSP)
   with a command line option '-d' for dot file generation,
   and added node/edge animation info in event numbers
   (one difference with other explorers like autexp(1): there
    the initial transition is named enil; here it is named e0)
   (and rewritten som stuff)
 * in primer(1): removed spurious debugging print statement
 * in primer(1): only complain about pattern in suspension
   when suspmode != recognize
 * in primer(1): when instantiating, treat quiescence as
   any other action, such that we can instantiate (e.g.)
   timestamps in it (so far quiescence was treated separately)
 * updated man page primer(1) with use of '*' in event pattern
 * updated man page primer(1) with allowed use of patterns
   in suspension if suspmode == recognize
 * (IN?)COMPATIBLE change: in primer(1) the matching
   of events with the patterns given in the configuration
   file is now done in a case insensitive way.
   (the change is actually in some 'common' (library) of
    code, there may be other pieces of TorX that use it too).

New in 3.8.1:
 * fixed configure script to have correct directory names
   of Src/Explorer/mcrl(2) directories
   (lowercase instead of uppercase)
 * in jararacy2anifsm be robust against a problem caused by
    an incompatible change in expect 5.42.1
 * added Explorer for CADP (so far we only had a Primer),
   mkprimer builds the CADP Primer by default
 * in CADP Explorer and Primer we now check
   MAX_FORMAT_LABEL before calling FORMAT_LABEL
   (to support open/caesar stuff that only supports basic format)
 * extended mkprimer witha --pmdirs flag to allow
   giving a list of directories in which to look for .pm files
 * visualization viewers no longer (send to) multicast step numbers
   for steps derived from their standard input
 * mctrl(1) now reads step numbers from standard input
   and uses these to update the scroll bar (slider)
 * new log2mctrl(1) derives step numbers from torx-log(4) file
 * anifsm(1) now allows simulataneous scrolling of clone-d
   windows: middle-click, or ctrl-middle-drag in one window
   makes all of its clones show the item under the mouse.
 * new torx_open(1) provides open/caesar interface to
   things that implement the torx-explorer(5) interface,
   so we can run the CADP simulator on them
 * INCOMPATIBLE change: new primexp(1) is a torx-explorer(5)
   that gives access to the LTS of a torx-primer(5).
   So, this allows us to simulate that LTS, e.g. using torx_open(1).
   NOTE: primexp used to be a shell script wrapper around
   primer(1) which is a torx-primer(5).

New in 3.8.0:
 * now really included the mkprimer support to connect
   with mCRL and mCRL2 already mentioned with but
   accidentally omitted from 3.7.0
* only install the mkprimer support for mCRL
   if both mcrl and mcrl2 toolkits are there
 * only install the mkprimer support for mCRL2
   if the mcrl2 toolkit is there
 * extended xtorx(1) with a Guides menu, like we already have
   Primers and Mutants menus
 * extended torx(1) with the ability to use the config file entries
   PRIMERS and PRIMER to select a primer, GUIDES and GUIDE
   to select a guide (test purpose), MUTANTS and MUTANT to
   select a mutant. Until now these config file entries were only
   used by xtorx(1), but now also torx(1) uses them if it is
   directly invoked by the user (if the --gui flag is not given).
 * minor fix to generate correct CONFIG INCLUDE torx-log(4)
   line from INCLUDE torx-config(4) entry.
 * changed autexp(1), jararaca(1), ltsa(1) to reuse state/event
   numbers used in their implementation of the torx-explorer(5)
   interface
 * fixed jararaca to allow just numbers with its torx-explorer(5)
   'e' command.
 * changed defaults for command-line options of jararaca(1)
   such that they are more like the other explorers: by default
   it now runs the explorer interface, by default it uses the
   action strings instead of the action 'symbols', and by default
   it now adds an epsilon selfloop on the accept state with label
   'EPSILON'
 * added additional command line options to jararaca(1)
   to override these new defaults
 * added mkprimer support for JARARACA primers where we
   do not add an epsilon selfloop to the accept state.
   Such jararaca files should have a .jrrc suffix;
   the suffix .tp is used for JARARACA test purpose files for
   which we do want the selfloop on the accept state.
 * added some documentation (mkprimer-aut(1))
 * extended log2anifsm(1) with -p command line option
   (like log2jararacy(1) already had)
 * log2anifsm(1) now shows the previous 'highlighting state'
   in grey (like log2jararacy(1) already did)

New in 3.7.0:
 * addded mkprimer support to connect with mCRL and mCRL2.
   For mCRL files this needs the mcrl command of the mcrl
   toolkit, and tbf2lpe and lpe2torx from the mcrl2 toolkit.
   For mCRL2 this needs the commands mcrl2lpe and lpe2torx
   from the mcrl2 toolkit.
 * added initial (experimental!) support for batch test generation.
   torx(1) generates a batch test of a specified depth in
   Aldebaran (.aut) format on its standard output if it (torx)
   is invoked with a --batch flag and the depth has been specified
   (either in the torx config (.if) file or via a --depth flag) --
   but only if the primer that is used supports interface commands
   STATEID and GOTO (see below)
   The result will be a tree by default; it will be a graph if
   the --batch-automaton flag is additionally given to torx(1).
 * extended the torx-primer(5) interface with a GOTO command
   to 'return' to a superstate which have visited before
   (i.e. GOTO gets as argument the ''super'' field we have
    obtained earlier using STATEID)
 * The GOTO command is now implemented in the CADP primer
    and in the 'generic' primer.
 * batch test generation is only partially supported by intersector(1):
   we can not yet support all cases in the cartesian product of
   {pass, fail} x {hit, miss}.
 * cleaned up autexp (and print some messages to standard error
   instead of standard output to avoid interference with other data
   written to standard output)
 * tried to slightly more improve state matching/numbering
   in primer(1) in the computation of field ``super'' of STATEID 
 * added command line options to log2aut(1) to choose which
   'features' (teststep, super state id, edge traversal count)
   are in the resulting automaton
 * in log2primer(1) output we include the ''channel'' in case
   of input or output suspension, because the primers to
   which we give the out output of log2primer(1) need it
 
New in 3.6.2:
 * extended anifsm to allow specification of the dot layout
   via stdin when special filename ``-'' is given
   (so the thing to animate can be extended ``on-the-fly'').
   This is intended for animation of the state space of the test run.
   NOTE: The syntax for this may change in the future --
   right now it is a bit different from the syntax in dot input file,
   to make it easier to parse. However, we may want to make it more
   identical to the dot file language -- so we may change it.
 * fix zooming in anifsm to not ``eat'' trailing ``0'' in node
   names when hiding and then showing again text during
   zooming out and then zooming in again
 * new log2anifsm outputs the extended anifsm input
 * new log2aut generates automaton for explored state space
 * new aut2fsmview translates .aut to input for FSMview tool
 * improved state matching/numbering scheme in primer(1) in
   computation of ``super'' field of STATEID 
   (so far it would also take into account the edge by which we
	reach the explorer states -- probably this codes dates from
	before we had edge information with the explorer states)
 * autexp now has a flag (-m) to write connectivity matrix
 * made parsing of Aldebaran (.aut) files more robust regarding
   white-space and quoting (single, double) of labels
 * added jararaca support for mkprimer
 * in the installation, use a separate script (subst) that now
   uses ``config.status'' to edit in configured variables
   into scripts etc.
   This could be _much_ slower though, in particular when the
   disk is not local, as when installing on a file server.
   (so far we directly used ``sed'' for that).
 * in the installation, catch error messages from tcl/expect
   commands to create tclIndex files etc.
 * tried to make installation more robust when kc is not available
   and unpacking messed up the time stamps of the files
 * tried to make installation more robust by having a dummy wish
   that tells you to install tk if you don't have wish (tk) installed.
 * to avoid line-length problems with that dummy wish, now we
   no longer start wish programs using ``#! /path/to/wish''
   but using ``#! /bin/sh'' and let the shell restart the script
   using wish.
 * removed a line-length limit from jararaca (so far, jararaca input
   should not contain lines longer than 255 characters)
 * tried to avoid/soften other parsing related limits in jararaca
 * fixed batch mode running of torx (would exit immediately on
   end-of-file on standard input, even when started in batch mode)
 * when running torx in batch mode, try to avoid unnecessarily starting
   (and leaving behind when torx dies) tmcs + logclient
 
New in 3.6.1:
 * updated README to be more explicit about flags for ./configure
 * once more work around some problems with older C compilers
   and tcl/tk interpreters, to be able to run ``everywhere''
 
New in 3.6.0:
 * mctrl now also gets installed, and has a man page
 * (ref explorer source actually not in distribution)
 * changed command language for anifsm(1) to be more general
   NOTE this is incompatible with the previous version;
   it is also incompatible with (unchanged) jararacy(1),
   so anifms's built-in fallback to jararacy (when tcldot not available)
   will no longer work. Actually, we should deprecate jararacy.
 * new jararacy2anifsm(1) translation program translates from old
   to new command language, and is used until we have a log2anifsm
   instead of (in addition to?) log2jararacy
 * changed Adapter: we now spawn a pty for each pco program
   to capture its stderr output separately (this seemed to be necessary
   to get it running (again) under cygwin)
 * added (HACK) new expnohup program to ignore HUP signals
   when anifsmsrv runs in the background
 * use INADDR_ANY in udp(1), so we can now use localhost
   instead of the `real' hostname in the examples
 * if in udp(1) reverse lookup fails for 127.0.0.1 we fall back to
   hard-coded name localhost (need seen under cygwin)
 * minor code cleanup in CADP Primer code (avoid compiler warnings)
 * updated CADP Primer code to be compatible with recent (post v2001)
   CADP (some CADP API changes had to be taken into account,
   without them the generated primer would just core dump)
 * in primer(1) we now return an error if the menu to choose
   from is empty ( e.g. GETINPUTERROR)
 * switched off debugging output in torx-stdiomerge
   TODO: as much as possible revert to not using torx-stdiomerge
   but instead spawing pty's to capture stderr separately from stdout
 * some minor bugfixes and silencing of warnings in the client/server-
   based visualization utilities (we should extract the client/server
   machinery to common and reuse that for all)
 
New in 3.5.5a:
 * intermediate release of timed stuff, also contains source,
   but no man, and no installation, for:
 * mctrl
 * explorer/ref (refiner)
 
New in 3.5.4:
* disable echo-ing from Driver commands sent to the Adapter
   (to avoid possible messed-up stderr output from Adapter or SUT)
* replaced the sh(1) version of torx-stdiomerge(1) by a perl version
  (alternative: get rid of torx-stdiomerge altogether, by reverting
   to the previous way of doing things, before 3.2.0)
* in xtorx(1) we now highlight the messages corresponding to the
  currrent test step (just like we highlight the current test step in
  the Path pane)
* avoid repeated identical error messages when we invoke xtorx(1)
  with a non-existing config file name
* for intersector(1) we reduce the number of messages
  when it fails to get a STATEID from a spec or guide
  it stops asking for it (to avoid repeatedly asking for it from a
  component that does not implement STATEID anyway)
* intersector(1) now tries to be better on matching actions
  that contain '* (to improve the dealing  withf wildcards)
* a Trojka generated Primer (see mkprimer-trojka(1)) the
  matching of `mtype' fields in actions received from the Driver
  is now case-insensitive. TODO: implement a warning to warn
  for mtype declarations that only differ in case (to avoid surprises)
* in anifsm(1), aniwait(1) and mscviewer we now use images
  on up/down buttons, instead of characters `+' and `-'
* added a -[no]delay option to udp(1) to simulate FIFO-buffering
  channels that incur delays

New in 3.5.3:
* try to please recent versions of automake
   (that complain about some of our Makefile.am files
    accepted by autmake 1.4) by fixing the things
   it warns about
 * added depcomp script (from automake 1.9.2),
   because relatively recent automake (1.7.8)  in recent red hat
   linux misses it, but does not add it by itself
   (unless 'missing' is also missing, then it does install depcomp)
 * updated the scripts missing, install-sh, mkinstalldirs,
   config.guess, config.sub, also from automake 1.9.2

New in 3.5.2:
* following fixes, enhancements, were not in local installations
   of 3.5.1 (so, real 3.5.2 work :-)
* reduced number of debug messages, warnings, etc.
* strict client/server separation for the visualization tools
   mscviewer, anifsm, aniwait, where the server no longer
   can be used as a client, but the client just tries to connect
   to a server, and if that fails, starts a server, and tries
   to connect to it
* tried to fix race conditions in startup of server part for
   mscviewer, anifsm, aniwait
* rewritten the clients for mscviewer, anifsm, aniwait in tcl,
  such that they no longer need external programs
* adaptsim now knows about the SOURCE config line (skips it)
* predicates related bugfixes in torx(1)
* fix in xtorx(1) to (again)  try to avoid resetting the toggle buttons
  in the Tools menu unintendedly when we reload primers/mutants
* tried (once again) to get the basic TorX stuff running under
   cygwin/windows, cygwin 1.5.11-1, works for basic (x)torx
   with .aut files (using autexp)
* portability fixes in the promela primer for windows with cygwin
   also a promela primer now seems to work under windows xp
   with cygwin 1.5.11-1
* now, when torx(1) fails to get a STATEID from a spec or an adapter
   it stops asking for it (to avoid repeatedly asking for it from a
   component that does not implement STATEID anyway)
* only pass predicates in C_INPUT and C_OUTPUT messages to
   Primer when non-empty.

New in 3.5.1:
* version 3.5.1 was an intermediate one (between 3.5.0 and 3.5.2)
  for local testing of the installation.
  Even though that version was not  ``frozen'' in the repository
  (so it is not an exact version) we have installed it on some
  machines. To avoid confusion of having multiple, different, 3.5.1
  versions, we just move on to 3.5.2.
* completed the `New in 3.5.0' (ahem)
* bugfix in trojka for a problem related to quiescense in combination
   of internal channels that could block
* new dot file animator anifsm(1)
   (like jararacy, but based on tcldot instead of on dotty), that
   has a -r command line option and associated toggle button,
   (to reuse the window)
* extended jararacy(1) to accept the same command line options
   as anifsm(1), such that we can use jararacy(1) as ``fall back''
   from anifsm(1), when tcldot is not available.
* new visualizer aniwait(1) for `waiting' during the time testing
* changed mscviewer(1) to look more like anifsm(1) and aniwait(1)
* bug fix in vertical scrolling of mscviewer(1)
* extended xtorx(1): the test run step currently shown by visualization
   tools like jararacy and mscviewer is highlighted, and can be changed
   using mouse Button-3
* when choosing the test step to visualize in the tools like
   jararacy(1), anifsm(1), aniwait(1), mscviewer(1) amd xtorx(1)
   we allow the step number chosen to be one higher than the
   latest exectuted test step, such that aniwait(1) can switch to the
   display of the remaining time for the current test step
* fix in torx(1) messages about INST-configured instantiator now
   refer to instantiator instead of `obusim'.

New in 3.5.0:
* bugfix in jararacy to avoid looping when multiple animators present
* new man page tmcs(1), to allow animation of the aut files using jararacy
* jararacy can now also highlight edges (not only nodes) while animating
* support in jararaca(1) to export the edge (transition) information as part
  of the state identifiers
* updated log2jararacy(1) to support this
* added `-p pattern' command line option to log2jararacy(1)
* new explorer autexp(1), to allow animation of the aut files using jararacy
   This aut explorer is now used by default by mkprimer; to use the old
   CADP-based aut explorer, invoke mkprimer with a flag: -l AUT-CADP
* quite some changes to support real time testing
  (still a bit experimental):
* new primer-driver interface command: C_GETOUTPUT
   (`symmetric' variant of C_OUTPUT, like C_GETINPUT for C_INPUT)
* C_GETOUTPUT is implemented by the generic symbolic primer(1),
   to allow adapters like adaptsim(1) that connect to a simulator,
   to ask for a symbolic output (observation) that can then be
   instantiated by a separate tool component, after which the
   instantiated result can be `committed' using the C_OUTPUT
   command.
* fixes in primer(1), to more completely support predicates
* fixes in primer(1), to better visualize edges for quiescent
   states and `identical' states
* change in primer(1), in `sim' mode, to choose the successor
   state from a set of possible destination states already when
   we do the `next' transition, such that the next `menu' is
   computed based on this choice, rather than `delaying' this
   decision and making the choice more implictly, when we
   choose a transition from the complete menu of all transistions
   offered by all possible destination states.
   This change offers a more interesting animation/visualization,
   because we now really see when the choice of a particular
   destination state cuts off (makes unreachable) certain succesor
   transitions.
* in adaptsim(1) more torx-primer(5) interface commands
  are implemented
* new config file (*.if) entries for the instantiator that can now be
  a `real' component: INST, INSTRUNDIR, INSTFLAGS,
  both for the driver and the adapter (adaptsim)
  (also this is a bit experimental, and has only been used for real-time
   experiments: the existing instantiator(1) has not yet been updated
   to use this interface)
* fixes in torx(1), when dealing with output from the adapter
* fixes in torx(1), to more completely support predicates
* new user-interface part in xtorx(1) to allow instantiation of
  predicates (initially hidden, enable via Preferences menu entry)

New in 3.4.1:
* bugfix in torx-logclient
  (to use blocking instead of non-blocking io)
  This improves performance a lot, and makes the
  msc viewer usable again.
* added window title code stuff for jararacy
  (planned for 3.4.0 but unfortuantely omitted)
* bugfix in jararacy, without this it could happen
  that the jararacy window would disappear
  'automatically' at the end of a test run. 

New in 3.4.0:
* View menu entries in xtorx(1) can be taken from
  the configuration file(s) (new SOURCE keyword)
* have slightly more informative window titles for
  msc and jararacy
  (using MUTANT and PRIMER configuration file entries)
* added tcp multicast service command tmcs(1)
  (TODO: add manual page for it)
* we now use tmcs(1) to allow a msc viewer and
  jararacy to update each others selection
  (if you step through an msc, then the selection
   in the jrararacy for the same test run will be
   changed too, and the other way around)
  POTENTIAL PROBLEM: it may be the case that tmcs
  processes are left running; please cleanup by
  hand (kill) when necessary. To Be Fixed.
  Note: there is likely still quite some debugging
  output for this. To Be Fixed.
* updated Trojka (Promela primer generator) to use
  stdarg.h instead of varargs.h in the generated code
  (some modern compilers removed support for varargs.h)
* slight extension to log2jararacy to deal with more
  complex (i.e. compound) states (hmm, was this for modest???)
* changed the log file format to have CONFIGSTART and CONFIGEND
  around the CONFIG lines (such that tools can detect the end
  of the CONFIG section, know no more CONFIG stuff will come,
  and start doing their thing). We now use this in log2msc.
* extended mscviewer(1) to make the start and end of msc
  update the selection too.

New in 3.3.2:
* fixed xtorx(1) to find the source files for
  primers and mutants (broken a while ago)
* in primer(1) we now printed slightly more warnings
  when an explorer reports that predicates were not solved

New in 3.3.1:
* fixed mscviewer(1) to keep the connection to Bmsc(1) open
  while files of the Bmsc(1) command are being processed,
  so mscviewer can report (syntax) errors via Bmsc.

New in 3.3.0:
* removed ``none'' entries from Primers and Mutants menu
* ``bugfix'' in torx-stdiomerge to deal with torx command line
  options for --log that contain a pipe (|) symbol.
  We need this in the confprot examples where we start gnuplot
  or jararacy. This is a HACK to make the example work;
  a real solution should be implemented someday.
  And has now be implemented (see below).
* added LOGMON configure keyword to torx, to start external
  tools that monitor (process) the text of the torx log file
  (see torx-config(4))
* added --logmon command line flag to torx (see torx(1))
  With this and LOGMON we have the real solution (see above)
  to no longer need torx --log command line options containing
  a pipe (|) symbol.
* added a (not to by used by the end user) torx-logclient(1)
  that is used in the implementation of LOGMON
* added a Tools menu to xtorx(1) which will contain a toggle
  button for each LOGMON entry in the config file(s), to
  enable/disable the starting of a log monitoring program
  before starting a test run
* implemented log2msc(1), mscviewer(1), Bmsc(1), xtorx-showmsc(1)
* removed integrated msc viewer from xtorx(1), now use
  xtorx-showmsc(1) via the --logmon option of torx(1)
* now indicate ``end of file''/``end of MSC'' in mscviewer.
* fixed minor bug in the mscviewer(1): it should no longer
  happen that parts of the labels in the MSC are invisible,
  to the left ``outside'' the MSC.
* the mscviewer(1) now has a ``Save as'' button which pops
  up a dialog window; if the filename selected/entered in
  the dialog ends in .(e)ps a postscript file will be saved,
  otherwise the ``event oriented'' textual representation
  of the MSC will be saved.
* fixed annoyance in the mscviewer(1):
  the ``instance headers'' now scroll upwards but remain
  visible, whereas they so far would just scroll off screen
* the mscviewer(1) now can highlight messages, and show
  the message (step) number, and has buttons to move the
  highlight up/down, and allows a step number to be
  entered, to highlight it and automatically scroll to
  make it visible
* the mscviewer(1) now has both a Close and a Quit button
* the mscviewer(1) now has a ``Reuse'' button to allow reuse
  of the window when a new MSC has to be displayed
* new ``Set MSC Window Reuse Button'' entry in xtorx
  preferences menu, to set the initial state of Reuse button
* now, during a test run the ``Start MSC'' menu entry,
  ``Set MSC Window Reuse Button'' and the new Tools menu/
  menu entries (see above) are disabled, because the new
  settings will not be taken into account during a test run,
  but only at the moment that a new test run is started.
* imnplemented an 'About' box in xtorx
* updated the xtorx(1) manual page
* bugfix in configure.in for configuring cpp for Trojka
* minor fixes in intersector, to better pass on suspension
  and channel information. The result of these fixes is that
  the log/msc info is more informative for experiments where
  the intersector is used, because the pco names appear in
  the log files and thus in the msc (previously, they would
  not be there if the intersector was used).
* (potential) fix in the tcp(.expect) program, in the handling
  of reading from stdin and (via the socket) from the peer
* added -w command line option to tcp(1)
* small fix in the tkgnats code (Src/third_party/tkgnats)
  to make it work with tcl/tk 8.4
* extended tkgnats code (Src/third_party/tkgnats)
  to be able to fill in the Release number automatically,
  so we now do that in torx-querypr and torx-sendpr
* fixed tcp(1) by disabling an event handler that would
  take up more and more cpu cycles
* added silly command torx-hostname(1) to (sort of) hack
  around host name confusion that we may have when we run
  in windows via cygwin; basically, it's only there to make
  the confprot example work in windows/cygwin
* found a problem when using a promela-based primer in
  windows/cygwin that crashed when the A_STATS answer
  tried to print the memusage which is a double.
  We don't understand why this happens; for now, we print
  a questionmark (?) for the memusage field in A_STATS.
* use tcl/tk logical font names to specify fonts for xtorx
  and mscviewer, so they look more the same (wrt font size)
  across platforms
* in xtorx the Primers, Mutants and Tools menu buttons are now no
  longer disabled during a test run (whether they are disabled
  or enabled now solely depends on whether or not a primers or
  mutants file is loaded, resp. whether there are entries in
  the Tools menu)
* The xtorx ``Path'' pane is now named ``Executed test steps''

New in 3.2.0:
* removed Trojka binary distribution support
* removed configure script flag --enable-trojka-src
* simplified ltsa configure support
* added configure script checks to avoid building/installing
  primer components for which support is not available
  in the build environment (e.g. missing interpreter)
* modified autoconfigure cpp path for trojka (needed for cygwin)
* ported most TorX components to cygwin
  (not ported: partitioner, iochooser, campaign)
* for cygwin, from tcl/tk/expect scripts, in many cases
  we now ``exec'' programs by explicitly calling
    exec /bin/sh -c "command line for the program"
* rewritten xtorx in tcl/tk, thus removed need for expectk
* added a --gui command line flag for torx
  (to make synchronisation in xtorx easier)
  Potential bug: the torx prompt ends with a space.
  If this space does not make it to xtorx, i.e. if
  xtorx sees the prompt but without the trailing space
  character, it will not recognize it.
* tried to make adaptor and adaptsim a bit more robust
* changed the way we deal with stdout and stderr output
  from TorX components (in some cases we now merge them
  and forward them over a single channel, instead of keeping
  them separate) with the result that in some cases the stderr
  output of a component will not appear to be independently
  continiously updated in xtorx
* extended the mkprimer interface to the *.pm primer-language
  dependent scripts to deal with the .exe suffix
* changed the configuration for generic Primer (primer(1)),
  both w.r.t. command line options and configuration file
* as a side effect, this primer now knows about suspension
  (quiescense) even when run in ``traces'' mode; therefore
  the intersector has been slightly modified
* fixed non-portability issue in CADP and Trojka Primer

New in 3.1.5:
* fixed buffer full problem in xtorx
* added mkprimer support for FSP (via ltsaexp(1))
* added --list and --config options to mkprimer
* now autoconfigure cpp path for trojka (needed for cygwin)

New in 3.1.4:
* fixed pco name extraction for msc display in xtorx
* minor textual updates to torx-primer(5) and torx-explorer(5) manual pages
* fixed expect regular expression patterns for end-of-line such
  that exp_internal debugging output of torx is now visible in xtorx.

New in 3.1.3:
* updates to some manual pages
* added PURL for torx web pages to torx-intro(1) man page
* added PURL and email contact address in README

New in 3.1.2:
* added a VERSION file, and updated LICENSE to reflect that

New in 3.1.1:
* bugfix in Adaptor to make old adaptor work again
  (needed for e.g. confprot example)

New in 3.1.0:
* Trojka now implements Tarjan algorithm
* integrated LICENSE file
* updated CONTRIBUTORS

New in 3.0.0:
* new primer-driver interface command: C_GETINPUT
* now primer does not do C_INPUT before the adapter does it,
  but after the adapter does it, where the adapter may change
  or extend the action that it is given (e.g. to fill in timestamp)
* lots of other things most certainly

New in 2.12.0:
* fixed the intersector to correctly handle commands un-implemented in the
  primers (etc.) that it controls
* added a preference menu option in xtorx to switch to bigger fonts
* changed the wording of '(Suspension)' back to '(Quiescence)'
* added a configure option to keep LTSA out of the distribution

New in 2.11.0:
* extended the generic primer with support for the ioct relation
  (in which the presence of tau transitions in a state is no reason
   to not add a delta transition). This support can be enabled with
 the -T flag of primexp
* added simple-minded support to allow the generic primer to do as if
  the thing it is exploring is input and or output complete,
  _not_ by extending the menu, but just by accepting an event
  that is not in the menu, telling that it has been accepted,
  and at the same time staying in the (same) super-state.
  (NOTE: this does not help if the current super-state contains
   more than one state, and the event that is asked for is in the
   menu for one the states, but not for all of them)
  This can be enabled with the primexp flag: -C input,output.
* fixed the generic primer: finally added the routine 'fatal'

New in 2.10.3:
* fixed the adaptsim adapter to receive the simulator stderr output separately
* fixed the generic primer: fixed the supension history heuristic for the
  IOKIND command so it will no longer forever avoid output once it has
  observed suspension
* fixed the generic primer: add 'identical state' information when a
  suspension transition is added to the 'menu' (without this fix we
  might not longer know that the current state has an 'identical' one,
  once we have gone through the 'suspension action' loop).
  In a way, we fixed a 'superstate-state-id' leak (without the fix
  the superstate-id number would grow and grow and grow)

New in 2.10.2:
* bugfix in the generic primer extension

New in 2.10.1:
* extended the generic primer such that we can also use it as simulator
  (at least for non-symbolic explorers), i.e. it implements also the
  'sim' 'strategy' (algorithm)

New in 2.10.0:
* integrated a new LTSA-based Explorer. It can be accessed in two ways:
  1) using the command 'ltsa' under the Check/Torx menu, which starts
  a telnet server on which the Explorer-Primer interface is offered, and
  2) using the command 'ltsaexp file.lts' which offers the
  Explorer-Primer interface on standard input and output.
  No manual pages yet.
  NOTE: LTSA needs java to run (javac to build from source, if provided).

New in 2.9.1:
* extended jararaca: we now also print the test-purpose expression
  in the dot file generated with jararaca's -r option
* extended jararaca: slightly optimised the generated automaton,
  by avoiding to unnecessarily increment a counter for the [n..]
  repeat operator (at the cost of adding one state and two
  transitions for each such operator)
* slightly rephrased part of the jararaca man page
* updated server name of problem report tracking system gnats
  in torx-sendpr and torx-querypr
* fix in msc window, to disallow 'Save in:' of empty msc window
  (which would fail with a tcl/tk error trace)

New in 2.9.0:
* extended jararaca:
  added a denotation for the 'number' 'infinity' in the recursion
  operator of the regular expression language: keyword: infinity

New in 2.8.4:
* changed STATEID answer in the primer-driver interface, to no
  longer contain the field 'all', and to always the field 'id'
  to 'super'
* updated the CADP primer, the generic symbolic primer and the
  intersector to handle the new interface
* updated the torx-primer.1 manual page w.r.t. this change

New in 2.8.3:
* bug fix in intersector (combinator) to complete the primer-driver
  interface code for STATEID, introduced in previous version
* added multi-lie STATETD answer to torx-primer.1 manual page

New in 2.8.2:
* now deal with STATEID and combinators in (sort of) the same way
  as we do with STATS: the intersector passes the STATEID info
  of its controlled primers etc. to its driver
* to do that, we extended the primer-driver interface for the STATEID
  command, to include the 'id,role' information
* changed the logging format for the STATEID lines: we now print
  them like the STATS lines as whitespace separated list
  of whitespace separated fields of name and value, and the first
  fields will be id and role, for tool components connected via
  a (or more) combinators (intersectors)
* bug fix in generic symbolic primer, to correctly give the superstate
  number representation (prevoiously, we always got the number '0')

New in 2.8.1:
* bug fix in intersector, to report
  'unknown [primer-driver interface] command' in the right way
  (such that the driver understands it)
* bug fix in driver, to remove superflous 'end of test' message
  that is given even before/while we are really starting

New in 2.8.0:
* new STATEID information in the log files generated by torx
* new STATEID command in the primer-driver interface,
  implemented in the CADP primer and the 'generic' primer
* bug fix to in the parsing of log files as traces, to make the
  follow-trace feature work again
* made jararaca input language accept optional ',' after
  last action in a set
* jararaca error messages now mention filename:lineno
* robustness improvement: torx (and other programs that use
  the io.tcl library for the primer-driver interface) can
  now better cope with long lines in the interface
* added manual pages: torx-explorer.1, smileexp.1, pui.1
* bug fix in generic symbolic primer: reimplemented its
  menu representation module (which was too complex, and
  contained and error which caused it to loose menu elements)
* bug fix in jararaca: removed the workaround that we needed
  to work around the error in the menu representation in the
  generic symbolic primer

New in 2.7.0:
* new jararaca explorer that derives traces from regular expressions
* bug fix in LOTOS and Promela primers, to write all interface
  messages, including error messages, to standard output,
  because in 2.6.0 we separated treatment in the driver of
  (explorer-) primer standard output and standard error
  (so, without this change we could deadlock becayse the driver
   would not see an A_ERROR message from the explorer/primer)
* extended configure script with a --with-litedir=... flag
  to allow specification of the path to the lite distribution
  (for tksmile, which can be used as symbolic explorer)
* slightly changed the generic symbolic primer to allow
  setting of environment variables to specify its explorer program,
  and the directory in which this explorer should be run.
* the gui (xtorx), now recognizes the (extended = including hit,miss)
  Verdict reported by torx, and displays it in the Verdict pane, and
  changes color accordingly
* small bugfixes in gui (xtorx), in particular to handle events
  coming from torx using the combinator (intersector) that do not
  contain channel or pco fields, but just the event 'label' and
  an indication of input or output. 

New in 2.6.0:
* bug fix in partitioner (error when C_INPUT was given with event)
* bug fix in campaign (requesting 'recursive' field of 'recursive'
  block types; e.g. requesting the 'partitioner' field of a
  partitioner block)
* bug fix in driver in torx user interface, allow ':' in channel+event
  argument of 'input' cmd (to allow copy-paste of output of menu cmd)
* bug fix in driver to support '$' symbols in events (used in free variables
  in symbolic events of the Smile explorer (in spec_get_input, where the
  (symbolic) event is used in a tcl list context)
* bugfix in instantiator to support  '$number' variable suffixes in
  the symbolic events of the smile explorer
* updated driver to capture and pass on stderr output of primer/explorer
* updated explorer-primer to capture and pass on stderr output of explorer
  (this symbolic primer is used with the smile explorer)
* added command (and man-page for) torx-root that prints the torx
  (distribution) installation directory
* bug fix in mksmileexp: generate call smileexp instead of utsmile
* minor tcl bug fix in smile explorer
* update msc displayer in xtorx, to automatically scroll the window
  when events are appended to the msc (i.e. to keep the 'end' visible)
* bugfix in iochooser: updated to 'new' tool interface where messages
  consist of list of name=value tuples

New in 2.5.1:
* bug fix in preprocmkprimer (for m4mkprimer and cppmkprimer)
* bug fix in driver and gui for logging to program instead of file

New in 2.5.0:
* bug fixes in generated campaign Makefiles to continue with next run
  when a torx execution ends with an error (instead of stopping campaign)
* updated combinator with the 'name-value tuple' interfaces 
* updated combinator (intersector) to latest theory
  (singular observation objectives; the existing prototype code for plural
   observation objectives has to be checked once theory is complete)
* when used with a combinator (intersector), torx's verdict now
  is of the form <{pass,fail}, {hit,miss}>.
* torx now also prints a verdict on stderr
* on fatal error in torx, if the last input/output command sent to
  the adapter was not successful, add this command to the log

New in 2.4.0:
* enhanced adapter to recognize LOG lines in stderr of implementation
* bug fixes in adapter to better handle eof of implementation
* extended driver to allow logging to a program instead of to a file
* extended campaign config with new fields
* bugfix in partitioner to correctly interface (use event and
  channel fields in the INPUT command)
* added a telnet-like script 'tcp' (like the script udp)
* removed some debugging print statements from adapter
* minor code cleanup in driver

New in 2.3.0:
* bug fixes in driver, adapter and hexcontext to give
  tcov-ed programs a chance to write out the coverage
  info when we stop

New in 2.2.1:
* verbosity flag (-v) added to partitioner command
* bug fixes in partitioner

New in 2.2.0:
* first version of tool component named partitioner, that can be
  use to partition the input menu taken from a 'sub' command (a primer,
  or another partitioner) using patterns over events (labels),
  and assign weights to the events using those partitions, such that
  it can select the next input using the weights.
* added partitioner support to the test-campaing code
* experimental primer user interface: pui
  usage: pui primer-program [arguments...]
* fixed some bugs (among other in the test-campaing .if file generator)

New in 2.1.3:
* bug fixes

New in 2.1.2:
* added milli-second timestamps (not very precise!)
* indicate the exeuction mode (auto/manual) in torx log 

New in 2.1.1:
* fixed handling of xtorx command line argument.
* fixed recognition of primer(s) and mutant(s) config file entries
  in adapter and driver (so they no longer complain about them)

New in 2.1.0:
* minor improvements to campaign support, to generate more entries
  in the generated torx config files, to allow xtorx to display
  more information
* minor improvements in xtorx, to make it easier to use settings from
  a test campaign directory, and to be able to specify primers and
  mutants files in the torx configuration file.
* xtorx now can be given one config file as argument
* a few bugfixes (esp. in the routine that compares events when
  a trace is followed)

New in 2.0.2:
* fixed distribution problem (Src/Explorer subdir Makefile* missing).

New in 2.0.1:
* Initial test campaign support (man page still missing).
* a number of bugfixes.

New in 2.0.0:
* Interfaces are more symmetric (p-d, d-a)
* separate adaptor module 
* added man pages
* configuration files extended 

New in 1.4.2:
* mkprimer now also generates a SOURCEREL setting in the
  generated outputfile.torx file.
* CHANGED the handling of the SOURCESPEC and SOURCEIUT
  keuwords in xtorx: if they refer to specs using relative
  paths, these relative paths are now interpreted relative
  to the location of the config file that contains it.
* we make now a clearer distinction in the output produced by
  torx: error/debug/diagnostic messages that need not be
  interpreted by xtorx (but should simply be appended to the
  message pane) are now send to standard error; only things
  that are ``important'' are send to standard output.
* a number of bugfixes.

New in 1.4.1:
* xtorx now has a Specs menu, with entries to open (view) the
  current spec or mutant (implementation).
* CHANGED torx now will by default always invoke a spec or iut
  such that the spawned subprocess will first 'cd' to the
  directory in which the spec/iut is, and then invoke the spec
  or iut. We have done this to make it easier to work with
  BCG files: primers that are build with bcg_io contain hard-
  coded pathnames that refer to the bcg file. The idea is that
  this change allows it to invoke mkprimer from within the
  directory containing the bcg file, with a relative path
  to the file (./spec.bcg or ./spec.aut) and then have it
  working correctly by default.
* CHANGED to overrule the default 'cd' behaviour mentioned
  above, two new config keywords have been introduced:
  SPECRUNDIR and IUTRUNDIR, which define the directory in
  which the SPEC resp. the IUT should be started;
  if the SPEC and/or the IUT program given has a relative
  path, then it is to be interpreted from SPECRUNDIR resp.
  IUTRUNDIR when these are given, otherwise they will be
  interpreted from the directory containing the config file.
  (POTENTIAL FUTURE CHANGE:
   we can also decide to interpret these
   program paths _always_ from the directory containing the
   config file, and just use the *RUNDIR config entries to
   specify where it has to be run)
* CHANGED the config file format: added two new keywords:
  SOURCESPEC and SOURCEIUT that can be used to specify
  where the specification resp. the spec or source of the
  iut can be found.
* mkprimer now always generates a outputfile.torx file
  containing a SOURCEABS setting, that can be used to
  locate the source, i.e. to set the SOURCESPEC or SOURCEIUT
  in a .primers or .mutants file.
  We also want to add a SOURCEREL setting, but it is not yet
  completely clear how to compute the relative path to a
  source file in the general case, i.e. to a file
  in a different subtree.
* mkprimer now has support to build a primer from an fc2 file
  (using bcg_io to read/translate the fc2 file)
* improved/generalised sim.primers/sim.mutants files in the
  Examples/Simulator directory, that use the .torx files.
* added a preference item to allow the user to say that the
  msc window should not be opened.
* a number of bugfixes.

New in 1.4.0:
* GCI support integrated in the torx distribution,
  in Src/EnDeCoding/GciGeneric.
* added the adapter for the primer/simulator to the torx distribution,
  in Src/EnDeCoding/GciSimulator.
  This directory has been set-up such that it can be build/installed
  as part of TorX, and also separately. So, hopefully, it can be used
  as a kind of example of how an adapter can be built for TorX, using
  the GciGeneric code.
* changed mkprimer script: the default primer name is now based on
  the specification file name, and this name can be overridden with
  a -o flag.
* Some fixes in the Makefiles to fix support for building outside
  the source directories.
* The GUI now has a Primers menu, just like for the mutants.
* The primers and mutants files can be reopened.
* CHANGED torx now accepts multiple config files on the command line.
* CHANGED the calling interface with which the primer/mutant
 'constructor' and 'destructor'  routines are called: they are now always
  invoked with two arguments: the names of the variables that can be set
  (using thc tcl upvar command)to add things to the command line
  of torx: the first variable is expanded at the end of the flags
  for torx, the second variable at the end of the config file arguments.

New in 1.3.3:
* torx now has new command line flags --no-log --no-trace --no-seed --no-depth
  to overrule the corresp. config file and use the default (in-program)
  values instead (--no-seed means: use default seed).
* torx now has a --help flag, and a more elaborate usage message
* xtorx now disables the seed entry during a test run (but this is almost
  invisible for the user: the entry does not look any different!)
* xtorx now invokes torx such that it will not use a log- or trace-file
  unless specified in the gui -- it will _always_ overrule the config
  file in this respect -- we still have to see how intuitive this is!
  It does this by using the new --no-log --no-trace --no-seed flags.

New in 1.3.2:
* xtorx now has a 'message' pane instead of a 'log' pane --
  all user-interface pieces related to it now refer to 'messages'
* xtorx now has menu-entries to set/unset the log-file name
  A chosen log filename is displayed in the window-title, and
  passed on to torx when torx is started.

New in 1.3.1:
* torx command-line flags now overrule config file entries
* xtorx now has a 'Seed' entry field; if non-empty when a test-run is
  started (i.e. when torx is started), the seed will be passed on to torx
  (if non-empty, it is supposed to be a number, but this is _not_ checked)

New in 1.3.0:
* new code in Intersector (should be renamed 'combinator')
  to be able to handle TGV-generated test-cases (and also test-graphs)
  by specifying them as a 'TEST' 'primer' in the config file of the
  intersector.
* addition to the mkprimer cadp.pm 'script module' to allow generation
  of a primer from a .aut file or a .bcg file
  (in case of the .aut file generated by TGV, it is necessary to rename
   the labels to get rid of TGV's input/output annotations, and to
   provide the verdict-information in a slightly different way;
   in case of a bcg file it should contain 'normal' LOTOS-like
   labels, because we don't have a script to rename the labels in/from
   a .bcg file. These renamings should be done _before_ mkprimer
   is called, i.e. mkprimer should be called on a .aut or .bcg file
   with nice 'LOTOS-like' labels!)
* fixed (worked around) installation problem with tkgnats lib dir creation

New in 1.2.1:
* symbolic stuff in Trojka
* GCI support in the Driver

New in 1.2.0:
* menubar
* integrated tkgnats to query and/or report problems (in Help menu)
* mutants menu (user-controlled), to allow easy choice of mutants

New in 1.1.4:
* Portability fixes for Linux and ecgs.
* Standard convention for distribution names.

New in 1.1.3:
* Portability fixes for Linux.

New in 1.1.2:
* Portability fixes for HP-UX. We no longer need nawk.

New in 1.1.1:
* Bugfix in the configure script. Slightly more robust scripts to invoke Trojka.

New in 1.1.0:
* Added Trojka to the distribution (in binary form, for sparc/solaris and i86/linux).

New in 1.0.4:
* Renamed ``tester'' to ``torx'' and ``xtester'' to ``xtorx''.
* Finally fixed the problem with ``make'' that caused some parts of the
  system not to be built.
* the ``auto-Refresh'' toggle is now by default enabled.

New in 1.0.3:
* The tester program now outputs all diagnostics from the IUT (instead of
  sometimes not showing the last lines before terminating)
* We now have a mkprimer program to build Primer programs for arbitrary
  specification languages (for now only supporting CADP to compile LOTOS).
* Now also tested (a bit) on Linux.

New in 1.0.1:
* It should now also install without CADP being available (LOTOS Primer will
  then not be installed

New in 1.0:
* First version to be distributed

New in 0.3 (or so):
* First version to taken outside UT for demo at Natlab