Estimated hours taken: 40
Branches: main
This is the second part of a change to support term dependency analysis
in the declarative debugger. A `mark' command is implemented for the
term browser, which allows a particular subterm to be selected and
returned from the browser. The declarative debugger interprets this as
a suspicious subterm, and tries to find a child or sibling node from which
this subterm comes. This is used to determine the next question to be
asked of the oracle.
browser/browse.m:
Update the browser interface to allow for marked subterms being
returned from the browser.
Implement and document the mark command.
Rewrite run_command as a switch instead of a chain of if-then-elses.
This forces all unimplemented commands to be explicitly listed,
and gives better error checking.
browser/browser_info.m:
Add a maybe_mark field to the browser_info. It is initially `no',
but is updated when the mark command is given.
browser/declarative_analyser.m:
Select which child or sibling node to ask about next by searching
for the origin of the suspicious subterm. If the subterm has mode
`out' we act as if the oracle had answered no, and if the subterm
has mode `in' we act as if the oracle had answered yes. In future
we may not wish to presume this -- we do so now mainly to keep the
analysis algorithm simpler.
browser/declarative_debugger.m:
Add a functor for suspicious subterms to the decl_answer type.
browser/declarative_oracle.m:
Accommodate the changed answer type. The oracle does not try to
store information about suspicious subterms in the knowledge base,
because in principle this could lead to infinite loops (although
currently this wouldn't be a problem since we don't ever use the
information to move upward in the tree, so no cycle could be
formed).
browser/declarative_user.m:
Accommodate the changed answer type, and interpret marked terms
from the browser as suspicious subterms.
browser/parse.m:
Add the new command.
browser/program_representation.m:
Add a procedure to convert the browser's list(dir) to a term_path.
Change atomic_goal_rep_is_call/2 so it fails for special predicates,
which was originally intended.
trace/mercury_trace_browse.c:
Ignore the extra argument -- marked terms are not currently used in
the main debugger.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/input_term_dep.*:
tests/debugger/declarative/output_term_dep.*:
tests/debugger/declarative/special_term_dep.*:
New test cases.
Estimated hours taken: 0.3
browser/declarative_user.m:
Remove the enclosing univ before browsing arguments in the
declarative debugger. This was the original intention.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/browse_arg.m:
tests/debugger/declarative/browse_arg.inp:
tests/debugger/declarative/browse_arg.exp:
A test case for this feature.
Estimated hours taken: 25
Enhance the configuration parameters of the browser. The browser is now
sensitive to the type of caller (print, browse, or print *) and can have
different settings for each type. E.g., the default line limit per variable
can be made shorter for `print *' than for `print'. The browser also allows
different settings for each format.
The browser also accepts an optional format, which overrides the current
setting for the duration of the call.
To make use of these features, mdb has a new `set' command that allows the
settings to be updated in various ways. The browser version of `set'
still works, although it doesn't yet accept the options that the former does.
Also, `print' and `browse' can now take options that override the default
format.
browser/browser_info.m:
New module which defines the key browser data structures. The
`browser_state' previously used by browse.m has been split into
two parts: the transient state (browser_info) and the persistent
state (browser_persistent_state). The persistent state is created
when the browser is first called, and is saved between calls to the
browser, whereas a new transient state is created each time the
browser is called, and lasts for the duration of the call.
The persistent state contains settings for each possible combination
of format and caller type.
This module exports C functions that update the persistent state.
browser/browse.m:
Move the browser_state into the new module.
Change the interface so that it allows optional format and caller
type input arguments. This allows its behaviour to vary depending on
whether the caller is implementing `print', `browse', or `print *'.
It also allows the default format to be overridden by options to
these commands.
General rearrangement of the browser code to avoid code duplication.
This addresses the following comment from the module:
% XXX: could abstract out the code common to the following preds.
along with other similar problems.
The code for testing the size of a term and deciding between
`io__write' and a depth limited printer has been moved from `print'
to `portray_flat'. This allows `print' to be used with formats
other than flat.
TODO: allow browser commands to take format options; currently only
mdb commands can.
browser/declarative_user.m:
Update this for the changed interfaces.
browser/mdb.m:
Add the new module.
browser/parse.m:
Move the definitions of `dir', `setting', and `portray_format'
to the interface of the new module.
The argument to the `<' command, which sets the depth limit, is
no longer optional. The default used to be to use the system
default, but now there are multiple system defaults and no way to
select which one.
browser/*.m:
tests/debugger/browse_pretty.inp:
tests/debugger/browse_pretty.exp:
s/clipx/width/g
s/clipy/lines/g
The reason for this change is that the names are used in the
context of all formats, but only verbose format actually clips
the output, so `clipx' and `clipy' are misleading.
tests/debugger/browser_test.inp:
tests/debugger/browser_test.exp:
Test the new features.
doc/user_guide.texi:
Document the new features.
trace/mercury_trace_browse.c:
trace/mercury_trace_browse.h:
Allow extra arguments for the print and browse functions, to specify
the format and the caller type (in the case of print).
Export the function MR_trace_set_browser_param, which calls the
browser to do the work of mdb's set command.
Export the types MR_Browse_Caller_Type and MR_Browse_Format, which
mirror two of the types in the new browser module.
trace/mercury_trace_external.c:
Update for the change to MR_trace_browse_one: pass `browse' as
the caller type, and always use the default format.
trace/mercury_trace_internal.c:
Add functions to parse the format and caller type options. Update
the print and browse commands to expect format options. Add the
`set' command.
Replace MR_trace_print_var and MR_trace_browse_var with one function,
MR_trace_browse_internal. This function takes the caller type as an
argument, so there is no need for two functions.
Remove occurrences of:
/* XXX should use MR_mdb_in and MR_mdb_out */
since we already do this.
trace/mercury_trace_vars.c:
trace/mercury_trace_vars.h:
Change the function type MR_Browser so that it also expects a caller
type and a format. These arguments are now required by the browser.
This means that some of the functions that take an MR_Browser
argument needed to be updated to also take a caller and/or format,
so they can pass this on to the browser function when it is called.
Estimated hours taken: 2.5
Distinguish between predicates and functions in the declarative debugger.
browser/declarative_execution.m:
Add a pred_or_func field to trace_atom.
trace/mercury_trace_declarative.c:
Construct trace atoms with the extra field.
browser/declarative_user.m:
Print function call results using function syntax.
browser/debugger_interface.m:
browser/util.m:
Move the definition of type pred_or_func to util.m, so it can
be used by the declarative debugger as well as the external debugger.
runtime/mercury_stack_layout.h:
Update a reference to the location of type pred_or_func.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/func_call.{m,inp,exp}:
Test case for the new feature.
tests/debugger/declarative/*.exp:
tests/debugger/declarative/*.exp2:
Update expected output from tests.
Estimated hours taken: 8
Use the term browser to print arguments in the declarative debugger.
browser/browse.m:
Remove (unused) io__state arguments from browse__init_state.
Export term_size_left_from_max/3 for use by the declarative
debugger.
browser/declarative_user.m:
Add a browser_state field to the user_state type.
Implement the `browse' command for checking questions as well as
for verifying the bug that is reported at the end of the process.
The original plan was to browse the whole atom at once, but this
is not possible with the current browser, so we instead browse one
argument at a time. The `browse' command now expects an integer
that specifies which argument.
Modify write_decl_atom/5 so that it checks whether the atom size
is below a certain limit. If not, then it calls browse__print/5
to do the output.
tests/debugger/declarative/app.exp:
tests/debugger/declarative/app.exp2:
tests/debugger/declarative/filter.exp:
tests/debugger/declarative/filter.exp2:
Update test cases.
Estimated hours taken: 8
Implement declarative debugging of code that throws exceptions.
This aborts if used on code that handles exceptions. There is no point
dealing with this yet, since the oracle won't handle the higher order
argument to try/2.
browser/declarative_debugger.m:
Add "unexpected exception" questions and "unhandled exception" bugs,
and generate these from EXCP events. Handle the exception case
in various switches.
Add a predicate unexpected_exception_children/4, analogous to
{wrong,missing}_answer_children/4.
browser/declarative_execution.m:
Add excp/5 nodes to the event trace, and export a C function to
construct them. Handle these nodes in various switches.
Allow contours to extend beyond NEGE events, if the status is
`undecided'. Such events have no matching NEGS or NEGF event,
so they do not mark the boundary of a separate context.
browser/declarative_oracle.m:
Store information about which exceptions should/shouldn't be thrown
from various calls, and use this information to answer questions
where possible.
browser/declarative_user.m:
Handle the new questions and bugs.
trace/mercury_trace_declarative.c:
Add a function to deal with EXCP events.
trace/mercury_trace_internal.c:
Allow declarative debugging to be started from EXCP events.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/throw.m:
tests/debugger/declarative/throw.inp:
tests/debugger/declarative/throw.exp:
tests/debugger/declarative/throw.exp2:
A test case for this feature.
tests/debugger/declarative/queens.exp:
Update the output from this test.
Estimated hours taken: 40
Implement the expanding of implicit subtrees in the declarative debugger.
When the maximum depth is reached by the front end, it now returns to
the back end a request for the missing subtree. If the back end receives
such a request, it restarts declarative debugging with a different
topmost call and a deeper depth bound.
The EDT instance needs to know when to request expansion, so CALL nodes
need a flag to indicate whether they were at the maximum depth. The
front end needs to be able to point out the bug and/or subtree to the
back end, so CALL, EXIT and FAIL nodes need to record the event number.
browser/declarative_execution.m:
- Store the event number in CALL, EXIT and FAIL nodes.
- Store a bool in CALL nodes which indicates whether the event
was at the maximum depth or not.
browser/declarative_debugger.m:
- Store the event number of the buggy event in the reported bug,
and pass this event number to the back end so it can go back
to that event.
- Add a case for expanding an implicit tree to the
diagnoser_response type, and handle this response properly.
- Export procedures to C that allow acces to the diagnoser_response
type.
- Accommodate the changes to the trace_node type.
browser/declarative_analyser.m:
- Store the list of previous prime suspects in the analyser state.
That way they don't have to be specially dealt with when
restarting analysis with an expanded subtree.
- When starting analysis, assume the top node is wrong; this
is not an unreasonable assumption, and the strategy works better
for the case when a subtree is expanded.
browser/declarative_user.m:
- Accommodate changes to the reported bug.
trace/mercury_trace_declarative.c:
- Change the depth step size to a reasonable number, now that
it works. This also has the effect of testing the change,
since some test cases go deeper than the new limit.
- Filter events outside the topmost call. Rather than keep
track of the minimum depth, we record the topmost call sequence
number and use a global to keep track of whether we have entered
or left this procedure.
- Factor out code in the existing mechanism for starting
declarative debugging, so that it can be used to re-start
debugging as well.
- Accommodate the changes to the trace_node type.
- Output error messages if declarative debugging fails to start
properly.
- Handle the reponse from the diagnoser, by jumping to the buggy
event (if a bug is found) or re-executing to expand a subtree
(if one is requested).
- Add a new checkpoint for events which are filtered out of
the annotated trace.
trace/mercury_trace_internal.c:
- Don't report error messages when declarative debugging fails
to start. Errors are now reported by the declarative debugger
before returning.
tests/debugger/declarative/*.inp:
tests/debugger/declarative/*.exp:
tests/debugger/declarative/*.exp2:
- Update to reflect the removed questions.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/filter.m:
tests/debugger/declarative/filter.inp:
tests/debugger/declarative/filter.exp:
- New test case to cover the code which filters events which
are outside the topmost call.
Estimated hours taken: 10
Represent bugs directly rather than with edt nodes. This makes the
interface to the analyser more consistent: bugs are now handled
in much the same way as questions.
browser/declarative_analyser.m:
- In the mercury_edt/2 typeclass, change edt_root/3 to
edt_root_question/3 and add the method edt_root_e_bug/3.
- Add the type prime_suspect/1 which keeps track of the
evidence (oracle answers) which implicates a particular
edt node.
- Use the new bug representation.
browser/declarative_debugger.m:
browser/declarative_oracle.m:
- Move oracle_confirmation to declarative_debugger.m, and
rename it decl_confirmation.
browser/declarative_debugger.m:
- Change decl_bug/1 to decl_bug/0. Instead of a bug being
represented by an edt node, decl_bug now represents
directly the different sorts of bugs possible.
- Add an implementation for the new mercury_edt method.
browser/declarative_oracle.m:
browser/declarative_user.m:
- Update the confirmation to use the new types.
tests/debugger/declarative/*.{exp,exp2}:
- Update test results.
Estimated hours taken: 15
Fix the user interface of the declarative debugger so that it
produces more readable output; the representation of atoms
is updated so that it can handle non-ground arguments.
Ensure that all DD output goes through the correct stream, so that
the debugger works properly under emacs.
browser/declarative_debugger.m:
Call a procedure in the oracle to handle bug confirmation
(including printing out the bug), rather than handle
it directly.
browser/declarative_execution.m:
Update the trace_atom type so that it can handle both ground
and free arguments. Export to C some procedures for
constructing trace atoms.
browser/declarative_oracle.m:
Export a procedure which handles bug confirmation. This calls
the declarative_user module to do the interaction, and interprets
the result.
browser/declarative_user.m:
Export a procedure to handle bug confirmation by the user.
Update to handle the changes to trace_atom.
trace/mercury_trace_declarative.c:
Construct trace_atoms by calling the new Mercury procedures
exported from browser/declarative_execution.m.
trace/mercury_trace_declarative.h:
Remove the macro that had been used to construct old style atoms.
trace/mercury_trace_vars.{c,h}:
Export a procedure to calculate the argument position of a
head variable.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/args.{m,inp,exp,exp2}:
New test case to test mixing bound and free arguments.
tests/debugger/declarative/*.inp:
tests/debugger/declarative/*.exp:
tests/debugger/declarative/*.exp2:
Update test cases to reflect new output, and the extra question
asked for bug confirmation.
Estimated hours taken: 1
Make all the modules in the browser library sub-modules of
module `mdb', to avoid link errors when users use module names
such as `parse'.
browser/Mmakefile:
browser/browser_library.m:
browser/mdb.m:
Rename browser_library.m to mdb.m.
Change `:- import_module' declarations to
`:- include_module' declarations.
browser/Mmakefile:
Remove the special case rule for `mer_browser.init' --
it doesn't work when the file names are not the same
as the module name. Instead, the default rule for `mdb.init'
is used and the output is copied to `mer_browser.init'.
browser/.cvsignore:
Rename header files, etc.
browser/*.m:
Add a `mdb__' prefix to the names of modules in the browser library
in `:- module' and `:- import_module' declarations.
trace/*.c:
Rename the header files for the browser library in
`#include' statements.
tests/hard_coded/Mmakefile:
tests/hard_coded/parse.m:
tests/hard_coded/parse.exp:
Test case.
Estimated hours taken: 120
Implement the new architecture for the front end of the
declarative debugger. Disable declarative debugging if conservative
GC is not being used since this change relies on it. The benefits
of declarative debugging in non-GC grades are not worth the extra
maintenance required to support it, since even small examples require
a large amount of memory.
browser/declarative_analyser.m:
New module which handles the analysis of EDTs, which was
previously done in browser/declarative_debugger.m.
browser/declarative_debugger.m:
- Change the interface according to the new design.
- Export types of the form decl_*, rather than edt_*.
- Move the EDT analysis to the new module.
- Handle the interaction between the analyser and the oracle.
- Don't interact directly with the user, but go via the oracle.
- New diagnoser state type.
- Update the EDT instance to conform to the other changes.
browser/declarative_oracle.m:
- Replace oracle_answer type with oracle_response.
- Allow the oracle to work with a queue of questions rather
than one at a time.
- Add new predicates to manipulate the knowledge base.
browser/declarative_user.m:
- Export the user_state type.
- Handle some user commands in this module.
- Handle questions properly (apart from printing atoms).
browser/browser_library.m:
Add the new module.
trace/mercury_trace_declarative.c:
- Initialise the front end state just once per session.
- Make the step size much larger, since implicit subtrees
are not fully supported yet.
- Correct the types in a format specifier.
trace/mercury_trace_declarative.c:
trace/mercury_trace_internal.c:
trace/mercury_trace_internal.h:
- Disable declarative debugging unless conservative GC is used.
Estimated hours taken: 160
Implement a new data structure for declarative debugging. The
major differences between this and the old implementation are:
- The data structure is implemented in Mercury. The definition
of the type, and procedures for constructing values of that
type, have been moved from trace/mercury_trace_declarative.{c,h}
to browser/declarative_execution.m (which is a new module).
- The front end no longer needs to call the back end via an
indirect pointer---the front end does not call the back end at
all.
- The data structure is not specifically for wrong answer
analysis, it is intended to be used for any sort of analysis.
- The data structure represents execution at a lower level---the
new front end defines a more abstract view in terms of this
data structure.
Implement a test harness for debugging the front end code. This allows
the front end to run as a stand-alone program, which can then be
debugged using `mdb'.
The code in the front end does not currently handle the new structure
very nicely. This is because that code is about to undergo some major
structural changes, so there is little point cleaning it up now.
Consequently:
- Some of the code in the front end is incorrect (eg. the
user interface does not print missing answer nodes
properly).
- The tests have not been reinstated.
These things will be fixed in subsequent changes.
Likewise the compiler still reserves two stack slots, even though
only one is required. After this change the algorithm should be able
to get away with using no stack slots, so modifications to the compiler
will be postponed until then.
browser/declarative_execution.m:
New module. Implement the execution_tree typeclass, which
represents the execution of a Mercury program. Implement
two instances of this typeclass, one for normal use and one for
testing purposes.
browser/declarative_test.m:
New module. A test harness that can be compiled as a
stand-alone program, enabling the front end to be debugged.
trace/mercury_trace_declarative.c:
trace/mercury_trace_declarative.h:
- Remove the definition of the old data structure.
- Add some macros which enable the new Mercury data structure
to be destructively updated by C code.
- Change the interface to this module so that it reflects more
general diagnosis, not just wrong answer analysis.
- Implement the new algorithm.
- Call an alternative front end if in test mode.
- Update comments.
trace/mercury_trace_internal.h:
Add a new mode for debugging the declarative debugger.
trace/mercury_trace_internal.c:
Change the command from `dd_wrong' to `dd', since it is not
specifically for wrong answer analysis. Add a new command
`dd_dd' which calls the alternative front end used for testing.
runtime/mercury_init.h:
runtime/mercury_wrapper.c:
runtime/mercury_wrapper.h:
util/mkinit.c:
Remove any reference to `MR_edt_root_node', since it is no
longer used.
browser/declarative_debugger.m:
- Add a case for missing answer nodes to `edt_node'.
- Change the `evaluation_tree' typeclass into `mercury_edt'
typeclass. This is to make it more distinct from the new
`execution_tree' typeclass, which is a lower level concept.
- New interface to the diagnoser: types `diagnoser_response'
and `diagnoser_state', and procedure `diagnosis'.
- Define an instance of `mercury_edt' from an instance of
`execution_tree'.
- Updates to the analyser to get it to compile---further changes
will be forthcoming.
browser/declarative_user.m:
- Updates to the user interface to get it to compile---further
changes will be forthcoming.
browser/browser_library.m:
Import the new module (declarative_execution.m).
browser/debugger_interface.m:
browser/util.m:
Move the definitions of trace_port_type and goal_path_string
to browser/util.m, since they are now used by more than just
the external debugger.
browser/Mmakefile:
Add the test harness as a `depend' target.
browser/browse_test.m:
Use the correct interface to the browser.
Estimated hours taken: 55
This change adds a more sophisticated oracle to the declarative
debugger. The oracle now tries to remember previous answers and uses
them if available, and only asks the user if there is no other
information.
browser/declarative_oracle.m:
- Add the type "oracle_assumption", which identifies an
internal assumption of the oracle.
- Add the type "oracle_answer", which the oracle returns to the
caller. This holds either the truth value, or a reason why
the truth value cannot be given yet.
- Implement a knowledge base to store the previous answers.
This data structure is designed to be able to store arbitrary
assertions, although at this stage only one kind of assertion
is used.
- Where possible, use the KB to answer questions, instead of
asking the user.
- Update comments.
browser/declarative_debugger.m:
- Use the new interface to the oracle. Handle "don't know"
answers in a simple way: either save the question and ask it
later or return without finding any bug.
- Move the debugger_command type to this module.
- Move the UI stuff to a new module.
browser/declarative_user.m:
New module to handle interaction between the debugger/oracle
and the user.
tests/debugger/declarative/Mmakefile:
tests/debugger/declarative/oracle_db.{m,inp,exp}
Add a test case for the new feature.
