state
Major Section: OTHER
Examples: (+ (@ y) 1) (assign a (aset1 'ascii-map-array (@ a) 66 'Upper-case-B))whereGeneral Form: (@ symbol)
symbol
is any symbol to which you have assign
ed a global
value. This macro expands into (f-get-global 'symbol state)
, which
retrieves the stored value of the symbol.
The macro assign
makes it convenient to set the value of a symbol.
The :
ubt
operation has no effect on the global-table
of state
.
Thus, you may use these globals to hang onto useful data structures
even though you may undo back past where you computed and saved
them.
Major Section: OTHER
Example Forms: (table acl2-defaults-table :defun-mode) ; current default defun-mode (table acl2-defaults-table :defun-mode :program) ; set default defun-mode to :program
See table for a discussion of tables in general. The legal keys for this table are shown below. They may be accessed and changed via the general mechanisms provided by tables. However, there are often more convenient ways to access and/or change the defaults. (See also the note below.)
:defun-modethe default defun-mode, which must be
:
program
or :
logic
.
See defun-mode for a general discussion of defun-modes. The
:
defun-mode
key may be conveniently set by keyword commands
naming the new defun-mode, :
program
and :
logic
.
See program and see logic.
:verify-guards-eagernessan integer between 0 and 2 indicating how eager the system is to verify the guards of a defun event. See set-verify-guards-eagerness.
:compile-fnsWhen this key's value is
t
, functions are compiled when they are
defun
'd; otherwise, the value is nil
. To set the flag,
see set-compile-fns.
:measure-functionthe default measure function used by
defun
when no :measure
is
supplied in xargs
. The default measure function must be a function
symbol of one argument. Let mfn
be the default measure function and
suppose no :measure
is supplied with some recursive function
definition. Then defun
finds the first formal, var
, that is tested
along every branch and changed in each recursive call. The system
then ``guesses'' that (mfn var)
is the :measure
for that defun
.
:well-founded-relationthe default well-founded relation used by
defun
when no
:
well-founded-relation
is supplied in xargs
. The default
well-founded relation must be a function symbol, rel
, of two
arguments about which a :
well-founded-relation
rule has been
proved. See well-founded-relation.
:invisible-fns-alistan alist used in the control of permutative rewriting. Each entry in this alist must be of the form
(fn ufn1 ufn2 ... ufnk)
, where fn
and each ufni
is a function symbol and the ufni
are all unary
functions. Such an entry makes each of the ufni
``invisible'' when
ordering the arguments of fn
via commutative
rules. See set-invisible-fns-alist.
:irrelevant-formals-okWhen this key's value is
t
, the check for irrelevant formals is
bypassed; otherwise, the value is the keyword nil
(the default).
See irrelevant-formals and see set-irrelevant-formals-ok.
:ignore-okWhen this key's value is
t
, the check for ignored variables is
bypassed; otherwise, the value is the keyword nil
(the default).
See set-ignore-ok.
:inhibit-warningsACL2 prints warnings that may, from time to time, seem excessive to experienced users. Each warning is ``labeled'' with a string identifying the type of warning. Consider for example
ACL2 Warning [Use] in ( THM ...): It is unusual to :USE ....Here, the label is "Use". The value of the key
:inhibit-warnings
is a list of such labels, where case is
ignored. Any warning whose label is a member of this list (where
again, case is ignored) is suppressed.
See set-inhibit-warnings and also
see set-inhibit-output-lst.
:bdd-constructorsThis key's value is a list of function symbols used to define the notion of ``BDD normal form.'' See bdd-algorithm and see hints.
:state-okThis key's value is either t or nil and indicates whether the user is aware of the syntactic restrictions on the variable symbol
STATE
.
See set-state-ok.
Note: Unlike all other tables, acl2-defaults-table
can affect the
soundness of the system. The table mechanism therefore enforces on
it a restriction not imposed on other tables: when table
is used to
update the acl2-defaults-table
, the key and value must be
variable-free forms. Thus, while
(table acl2-defaults-table :defun-mode :program),are all examples of legal events (assuming(table acl2-defaults-table :defun-mode ':program), and
(table acl2-defaults-table :defun-mode (compute-mode *my-data*))
compute-mode
is a
function of one non-state
argument that produces a defun-mode as
its single value),
(table acl2-defaults-table :defun-mode (compute-mode (w state)))is not legal because the value form is
state
-sensitive.Consider for example the following three events which one might make into the text of a book.
(in-package "ACL2")The second event is illegal because its value form is(table acl2-defaults-table :defun-mode (if (ld-skip-proofsp state) :logic :program))
(defun crash-and-burn (x) (car x))
state
-sensitive. If it were not illegal, then it would set the
:
defun-mode
to :
program
when the book was being certified but
would set the defun-mode to :
logic
when the book was being loaded
by include-book
. That is because during certification,
ld-skip-proofsp
is nil
(proof obligations are generated and
proved), but during book inclusion ld-skip-proofsp
is non-nil
(those obligations are assumed to have been satisfied.) Thus, the
above book, when loaded, would create a function in :
logic
mode that
does not actually meet the conditions for such status.
For similar reasons, table
events affecting acl2-defaults-table
are
illegal within the scope of local
forms. That is, the text
(in-package "ACL2")is illegal because(local (table acl2-defaults-table :defun-mode :program))
(defun crash-and-burn (x) (car x))
acl2-defaults-table
is changed locally. If
this text were acceptable as a book, then when the book was
certified, crash-and-burn
would be processed in :
program
mode,
but when the certified book was included later, crash-and-burn
would have :
logic
mode because the local
event would be skipped.The text
(in-package "ACL2")is acceptable and defines(program) ;which is (table acl2-defaults-table :defun-mode :program)
(defun crash-and-burn (x) (car x))
crash-and-burn
in :
program
mode, both
during certification and subsequent inclusion.
We conclude with an important observation about the relation between
acl2-defaults-table
and include-book
, certify-book
, and encapsulate
.
Including or certifying a book never has an effect on the
acl2-defaults-table
, nor does executing an encapsulate
event; we
always restore the value of this table as a final act. (Also
see include-book, see encapsulate, and
see certify-book.) That is, no matter how a book fiddles with
the acl2-defaults-table
, its value immediately after including that
book is the same as immediately before including that book. If you
want to set the acl2-defaults-table
in a way that persists, you need
to do so using commands that are not inside books. It may be useful
to set your favorite defaults in your acl2-customization
file;
see acl2-customization.
state
Major Section: OTHER
Examples: (assign x (expt 2 10)) (assign a (aset1 'ascii-map-array (@ a) 66 'Upper-case-B))whereGeneral Form: (assign symbol term)
symbol
is any symbol (with certain enforced exclusions to
avoid overwriting ACL2 system ``globals'') and term
is any ACL2
term that could be evaluated at the top-level. Assign
evaluates
the term, stores the result as the value of the given symbol in the
global-table
of state
, and returns the result. (Note: the
actual implementation of the storage of this value is much more
efficient than this discussion of the logic might suggest.)
Assign
is a macro that effectively expands to the more
complicated but understandable:
(pprogn (f-put-global 'symbol term state) (mv nil (f-get-global 'symbol term state) state)).The macro
@
gives convenient access to the value of such globals.
The :
ubt
operation has no effect on the global-table
of state
.
Thus, you may use these globals to hang onto useful data structures
even though you may undo back past where you computed and saved
them.
certify-book
Major Section: OTHER
Examples: (certify-book! "my-arith" 3) ;Certify in a world with 3 ; commands, starting in a world ; with at least 3 commands. (certify-book! "my-arith") ;Certify in the initial world. (certify-book! "my-arith" 0 nil) ;As above, but do not compile.whereGeneral Form: (certify-book! book-name k compile-flg)
book-name
is a book name (see book-name), k
is a
nonnegative integer used to indicate the ``certification world,''
and compile-flg
indicates whether you wish to compile the
(functions in the) book.
This command is identical to certify-book
, except that the second
argument k
may not be t
in certify-book!
and if k
exceeds the current command number, then an appropriate ubt!
will
be executed first. See certify-book and see ubt!.
Major Section: OTHER
Examples: :comp t ; compile all uncompiled ACL2 functions :comp foo ; compile the defined function foo :comp (foo bar) ; compile the defined functions foo and barNote: Unless a single function is specified (either as a symbol or as a one-element list containing its name), ACL2 will write out filesGeneral Form: :comp specifier where specifier is one of the following:
t compile all defined ACL2 functions that are currently uncompiled (name-1 ... name-k) a non-empty list of names of functions defined by DEFUN in ACL2 name same as (name)
"TMP.lisp"
and "TMP1.lisp"
(the latter is for the
executable counterparts; see executable-counterpart) that are
subsequently compiled. Also see set-compile-fns for a way to
get each function to be compiled as you go along.
Major Section: OTHER
Example: ACL2 !>:good-byeNote: Your entire session will disappear forever when you type
:good-bye
.
The command :good-bye
quits not only out of the ACL2 command loop,
but in fact quits entirely out of the underlying Lisp. Thus, there
is no going back! You will not be able to re-enter the command loop
after typing :good-bye
! All your work will be lost!!!
This command may not work in some underlying Common Lisp implementations. But we don't expect there to be any harm in trying. It does work in AKCL and GCL, at least as of this writing.
If you merely want to exit the ACL2 command loop, use :
q
instead
(see q).
Major Section: OTHER
Example: (LD standard-oi ; open obj in channel, stringp file name ; to open and close, or list of forms; Optional keyword arguments: :standard-co ... ; open char out or file to open and close :proofs-co ... ; open char out or file to open and close :current-package ... ; known package name :ld-skip-proofsp ... ; nil, 'include-book, t, or ; see ld-skip-proofsp :ld-redefinition-action ; nil or '(:a . :b) :ld-prompt ... ; nil, t, or some prompt printer fn :ld-keyword-aliases ... ; an alist pairing keywords to parse info :ld-pre-eval-filter ... ; :all, :query, or some new name :ld-pre-eval-print ... ; nil, t, or :never :ld-post-eval-print ... ; nil, t, or :command-conventions :ld-evisc-tuple ... ; nil or '(alist nil nil level length) :ld-error-triples ... ; nil or t :ld-error-action ... ; :return (default), :continue or :error :ld-query-control-alist ; alist supplying default responses :ld-verbose ...) ; nil or t
Ld
is the top-level ACL2 read-eval-print loop. (When you call lp
, a
little initialization is done in raw Common Lisp and then ld
is
called.) ld
is also a general-purpose ACL2 file loader and a
command interpreter. Ld
is actually a macro that expands to a
function call involving state
. Ld
returns an ``error/value/state''
triple as explained below.
The arguments to ld
all happen to be global variables in state
. For
example, '
current-package
and '
ld-verbose
are global variables,
which may be accessed via (@ current-package)
and (@ ld-verbose)
.
When ld
is called, it ``binds'' these variables. By ``binds'' we
actually mean the variables are globally set but restored to their
old values on exit. Because ld
provides the illusion of state
global variables being bound, they are called ``ld
specials'' (after
the Lisp convention of calling a variable ``special'' if it is
referenced freely after having been bound).
Note that all arguments but the first are passed via keyword. Any
variable not explicitly given a value in a call retains its pre-call
value, with the exception of :
ld-error-action
, which defaults to
:return
if not explicitly specified.
Just as an example to drive the point home: If current-package
is
"ACL2"
and you typed
(ld *standard-oi* :current-package "MY-PKG")you would find yourself in (an inner) read-eval-print loop in which the current-package was
"MY-PKG"
. You could operate
there as long as you wished, changing the current package at will.
But when you typed :
q
you would return to the outer
read-eval-print loop where the current package would still be
"ACL2"
.
Roughly speaking, ld
repeatedly reads a form from standard-oi
,
evaluates it, and prints its result to standard-co
. It does this
until the form evaluates to an error triple whose value component is
:
q
or until the input channel or list is emptied. However, ld
has
many bells and whistles controlled by the ld
specials. Each such
special is documented individually. For example, see the
documentation for standard-oi
, current-package
,
ld-pre-eval-print
, etc.
A more precise description of ld
is as follows. In the description
below we use the ld
specials as variables, e.g., we say ``a form is
read from standard-oi
.'' By this usage we refer to the current
value of the named state global variable, e.g., we mean ``a form is
read from the current value of '
standard-oi
.'' This technicality has
an important implication: If while interacting with ld
you change
the value of one of the ld
specials, e.g., '
standard-oi
, you will
change the behavior of ld
, e.g., subsequent input will be taken from
the new value.
Three ld
specials are treated as channels: standard-oi
is treated as
an object input channel and is the source of forms evaluated by ld
;
standard-co
and proofs-co
are treated as character output channels
and various flavors of output are printed to them. However, the
supplied values of these specials need not actually be channels;
several special cases are recognized. If the supplied value of one
of these is in fact an open channel of the appropriate type, that
channel is used and is not closed by ld
. If the supplied value of
one of these specials is a string, the string is treated as a file
name and a channel of the appropriate type is opened to/from that
file. (Note that ld
does not support structured pathnames, as
described in the documentation for pathname
.) Any channel opened
by ld
during the binding of the ld
specials is automatically closed
by ld
upon termination. If standard-co
and proofs-co
are equal
strings, only one channel to that file is opened and is used for
both.
Several other alternatives are allowed for standard-oi
. If
standard-oi
is a true list then it is taken as the list of forms to
be processed. If standard-oi
is a list ending in an open channel,
then ld
processes the forms in the list and then reads and processes
the forms from the channel. Analogously, if standard-oi
is a list
ending a string, an object channel from the named file is opened and
ld
processes the forms in the list followed by the forms in the
file. That channel is closed upon termination of ld
.
The remaining ld
specials are handled more simply and generally have
to be bound to one of a finite number of tokens described in the
:
doc
entries for each ld
special. Should any ld
special be supplied
an inappropriate value, an error message is printed.
Next, if ld-verbose
is t
, ld
prints the message ``ACL2 loading
name'' where name
names the file or channel from which forms are
being read. At the conclusion of ld
, it will print ``Finished
loading name'' if ld-verbose
is t
.
Finally, ld
repeatedly executes the ACL2 read-eval-print step, which
may be described as follows. A prompt is printed to standard-co
if
ld-prompt
is non-nil
. The format of the prompt is determined by
ld-prompt
. If it is t
, the default ACL2 prompt is used. If it is
any other non-nil
value then it is treated as an ACL2 function that
will print the desired prompt. See ld-prompt. In the
exceptional case where ld
's input is coming from the terminal
(*standard-oi*)
but its output is going to a different sink (i.e.,
standard-co
is not *standard-co*
), we also print the prompt to the
terminal.
Ld
then reads a form from standard-oi
. If the object read is a
keyword, ld
constructs a ``keyword command form'' by possibly
reading several more objects. See keyword-commands. This
construction process is sensitive to the value of
ld-keyword-aliases
. See ld-keyword-aliases. Otherwise, the
object read is treated as the command form.
Ld
next decides whether to evaluate or skip this form, depending on
ld-pre-eval-filter
. Initially, the filter must be either :all
,
:query
, or a new name. If it is :all
, it means all forms are
evaluated. If it is :query
, it means each form that is read is
displayed and the user is queried. Otherwise, the filter is a name
and each form that is read is evaluated as long as the name remains
new, but if the name is ever introduced then no more forms are read
and ld
terminates. See ld-pre-eval-filter.
If the form is to be evaluated, first prints the form to
standard-co
, if ld-pre-eval-print
is t
. With this feature, ld
can
process an input file or form list and construct a script of the
session that appears as though each form was typed in.
See ld-pre-eval-print.
Ld
then evaluates the form, with state
bound to the current state.
The result is some list of multiple values. If a state is among the
values, then ld
uses that state as the subsequent current state.
Depending on ld-error-triples
, ld
may interpret the result as an
``error.'' See ld-error-triples. We first discuss ld
's
behavior if no error signal is detected (either because none was
sent or because ld
is ignoring them as per ld-error-triples
).
In the case of a non-erroneous result, ld
does two things: First, if
the logical world in the now current state is different than the
world before execution of the form, ld
adds to the world a ``command
landmark'' containing the form evaluated.
See command-descriptor. Second, ld
prints the result to
standard-co
, according to ld-post-eval-print
. If ld-post-eval-print
is nil
, no result is printed. If it is t
, all of the results are
printed as a list of multiple values. Otherwise, it is
:command-conventions
and only the non-erroneous ``value'' component
of the result is printed. See ld-post-eval-print.
Whenever ld
prints anything (whether the input form, a query, or
some results) it ``eviscerates'' it if ld-evisc-tuple
is non-nil
.
Essentially, evisceration is a generalization of Common Lisp's use
of *print-level*
and *print-length*
to hide large substructures.
See ld-evisc-tuple.
We now return to the case of a form whose evaluation signals an
error. In this case, ld
first restores the ACL2 logical world to
what it was just before the erroneous form was evaluated. Thus, a
form that partially changes the world (i.e., begins to store
properties) and then signals an error, has no effect on the world.
You may see this happen on commands that execute several events
(e.g., an encapsulate
or a progn
of several defuns
): even though the
output makes it appear that the initial events were executed, if an
error is signalled by a later event the entire block of events is
discarded.
After rolling back, ld
takes an action determined by
ld-error-action
. If the action is :continue
, ld
merely iterates the
read-eval-print step. Note that nothing suggestive of the value of
the ``erroneous'' form is printed. If the action is :return
, ld
terminates normally. If the action is :error
, ld
terminates
signalling an error to its caller. If its caller is in fact another
instance of ld
and that instance is watching out for error signals,
the entire world created by the erroneous inner ld
will be discarded
by the outer ld
.
Ld
returns an error triple, (mv erp val state)
. Erp
is t
or nil
indicating whether an error is being signalled. If no error is
signalled, val
is the ``reason'' ld
terminated and is one of :exit
(meaning :
q
was read), :eof
(meaning the input source was
exhausted), :error
(meaning an error occurred but has been
supressed) or :filter
(meaning the ld-pre-eval-filter
terminated
ld
).
Major Section: OTHER
Example: :props assoc-eq
Props
takes one argument, a symbol, and prints all of the properties
that are on that symbol in the ACL2 world.
:q
) -- reenter with (lp)
Major Section: OTHER
Example: ACL2 !>:Q
The keyword command :q
typed at the top-level of the ACL2 loop will
terminate the loop and return control to the Common Lisp top-level
(or, more precisely, to whatever program invoked lp
). To reenter
the ACL2 loop, execute (acl2::lp)
in Common Lisp. You will be in
the same state as you were when you exited with :q
, unless during
your stay in Common Lisp you messed the data structures
representating the ACL2 state.
Unlike all other keyword commands, typing :q
is not equivalent to
invoking the function q
. There is no function q
.
Major Section: OTHER
Examples: ACL2 !>(rebuild "project.lisp") ACL2 !>(rebuild "project.lisp" t) ACL2 !>(rebuild "project.lisp" :all) ACL2 !>(rebuild "project.lisp" :query) ACL2 !>(rebuild "project.lisp" 'lemma-22)
Rebuild
allows you to assume all the commands in a given file or
list, supplied in the first argument. Because rebuild
processes an
arbitrary sequence of commands with ld-skip-proofsp
t
, it is
unsound! However, if each of these commands is in fact admissible,
then processing them with rebuild
will construct the same logical
state that you would be in if you typed each command and waited
through the proofs again. Thus, rebuild
is a way to reconstruct a
state previously obtained by proving your way through the commands.
The second, optional argument to rebuild
is a ``filter''
(see ld-pre-eval-filter) that lets you specify which commands
to process. You may specify t
, :all
, :query
, or a new logical name.
t
and :all
both mean that you expect the entire file or list to be
processed. :query
means that you will be asked about each command
in turn. A new name means that all commands will be processed as
long as the name is new, i.e., rebuild
will stop processing commands
immediately after executing a command that introduces name. Rebuild
will also stop if any command causes an error. You may therefore
wish to plant an erroneous form in the file, e.g., (mv t nil state)
,
(see ld-error-triples), to cause rebuild
to stop there. The
form (i-am-here)
is such a pre-defined form. If you do not specify
a filter, rebuild
will query you for one.
Inspection of the definition of rebuild
, e.g., via :
pc
rebuild-fn
,
will reveal that it is just a glorified call to the function ld
.
See ld if you find yourself wishing that rebuild
had additional
functionality.
ld
specials
Major Section: OTHER
Examples: (reset-ld-specials t) (reset-ld-specials nil)
Roughly speaking, the ld
specials are certain state global
variables, such as current-package
, ld-prompt
, and
ld-pre-eval-filter
, which are managed by ld
as though they were
local variables. These variables determine the channels on which ld
reads and prints and control many options of ld
. See ld for
the details on what the ld
specials are.
This function, reset-ld-specials
, takes one Boolean argument, flg
.
The function resets all of the ld
specials to their initial,
top-level values, except for the three channel variables,
standard-oi
, standard-co
, and proofs-co
, which are reset to their
initial values only if flg
is non-nil
. Of course, if you are in a
recursive call of ld
, then when you exit that call, the ld
specials
will be restored to the values they had at the time ld
was called
recursively. To see what the initial values are, inspect the value
of the constant *initial-ld-special-bindings*
.
Major Section: OTHER
The top-level ACL2 loop has a variable which controls which sense of
execution is provided. To turn ``guard checking on,'' by which we
mean that guards are checked at runtime, execute the top-level form
:set-guard-checking t
. To turn it off, do :set-guard-checking nil
.
The status of this variable is reflected in the prompt.
ACL2 !>means guard checking is on and
ACL2 >means guard checking is off. The exclamation mark can be thought of as ``barring'' certain computations. The absence of the mark suggests the absence of error messages or unbarred access to the logical axioms. Thus, for example
ACL2 !>(car 'abc)will signal an error, while
ACL2 >(car 'abc)will return
nil
.
Whether guards are checked during evaluation is independent of the
default-defun-mode
. We note this simply because it is easy to
confuse ``:
program
mode'' with ``evaluation in Common Lisp'' and
thus with ``guard checking on;'' and it is easy to confuse
``:
logic
mode'' with ``evaluation in the logic'' and with ``guard
checking off.'' But the default-defun-mode
determines whether
newly submitted definitions introduce programs or add logical
axioms. That mode is independent of whether evaluation checks
guards or not. You can operate in :
logic
mode with runtime guard
checking on or off. Analogously, you can operate in :
program
mode with runtime guard checking on or off.
However, one caveat is appropriate: functions introduced only as
programs have no logical definitions and hence when they are
evaluated their Common Lisp definitions must be used and thus their
guards (if any) checked. That is, if you are defining
functions in :
program
mode and evaluating expressions
containing only such functions, guard checking may as well be
on because guards are checked regardless. This same caveat
applies to a few ACL2 system functions that take state
as an
argument. This list of functions includes all the keys of the alist
*super-defun-wart-alist*
and all function whose names are members
of the list *initial-untouchables*
.
See guard for a general discussion of guards.