Major Section: ACL2 Documentation
Call this up with (verify ...)
.
VERIFY
command
(see verify), and then invokes commands at the resulting prompt to operate
on a stack of goals, starting with the single goal that was supplied to
VERIFY
. The final command (or ``instruction'') can be an exit
command, which can print out a defthm
event if the goal stack is empty;
see proof-checker-commands, in particular the exit
command. That
resulting defthm
event includes an :
instructions
parameter,
which directs replay of the proof-checker commands (for example during
certification of a book containing that event; see books).
If you exit the proof-checker interactive loop, you may re-enter that session
at the same point using the command (verify)
, i.e., with no arguments.
The commands save
and retrieve
may be invoked to manage more than one
session.
The proof-checker can be invoked on a specific subgoal, and the resulting
:instructions
can be given as a hint to the theorem prover for that
subgoal. See instructions.
A tutorial is available on the world-wide web:
http://www.cs.utexas.edu/users/kaufmann/tutorial/rev3.html.
The tutorial illustrates more than just the proof-checker. The portion
relevant to the proof-checker may be accessed directly:
http://www.cs.utexas.edu/users/kaufmann/tutorial/rev3.html#slide29
See set-evisc-tuple for how to arrange that output is printed in abbreviated
form. In general, the proof-checker uses the :TERM
evisc-tuple
described in that documentation.
Individual proof-checker commands are documented in subsection
proof-checker-commands. When inside the interactive loop (i.e., after
executing verify
), you may use the help
command to get a list of
legal instructions and (help instr)
to get help for the instruction
instr
.