compiler/unify_proc.m:
When creating goals of the form
X = f(ArgX1, ArgX2),
Y = f(ArgY1, ArgY2),
...
don't traverse (a) the types of the arguments of f, (b) the ArgXi, and
(c) the ArgYi separately, living with the possibility that the lists
containing them have different lengths. Instead, create a list of triples,
and traverse that.
Give names such as ArgX1 to these variables. (They used to be unnamed.)
Reorder things to ensure that the order of predicate/function declarations
is matched by the order of their definitions.
tests/invalid/partial_implied_mode.err_exp:
Expect ArgX2 to be a named variable in an error message.
compiler/mode_errors.m:
When printing insts in error messages, use predicates specifically
designed for that purpose.
The insts output by these predicates improve on the existing output
in the following ways.
- We now show the structure of complex insts through indentation.
- We now show both the name of named insts, and their expansion.
Sometimes, you need the expanded inst to understand the error message,
in the case of recursive insts, you cannot expand them forever.
Since the expansion *has* to refer to the inst name, showing that name
at the *start* of the expansion is necessary to allow readers to
understand the inst.
- We now eliminate the module qualification from the names of function
symbols. The module qualifiers on function symbols are specified
by the type, not the inst, of the relevant term, and if we get to
the point of mode analysis having been run, then the predicate
we are generating errors for must be *type* correct, though obviously
it is not *mode* correct.
- We now eliminate the module qualification from names of named insts
if the module qualifier is the name of the module being compiled.
This is a common case, and in these cases, the module qualifier
in the error message acts as clutter, making it *harder* to read.
- We now avoid printing parentheses that humans never need, but which
the parser sometimes does, and which the existing inst output predicates
therefore had to print.
- Higher order insts contain the modes of the function or predicate.
We now print the builtin modes as e.g. "in" instead of the old
"free >> ground".
compiler/error_util.m:
Add some utility predicates now needed by mode_errors.m.
tests/invalid/any_passed_as_ground.err_exp2:
tests/invalid/anys_in_negated_contexts.err_exp:
tests/invalid/bind_in_negated.err_exp:
tests/invalid/bug117.err_exp:
tests/invalid/bug191.err_exp:
tests/invalid/constrained_poly_insts.err_exp:
tests/invalid/constrained_poly_insts2.err_exp:
tests/invalid/ho_any_inst.err_exp:
tests/invalid/ho_default_func_1.err_exp:
tests/invalid/ho_default_func_2.sub.err_exp:
tests/invalid/ho_default_func_3.err_exp:
tests/invalid/ho_default_func_4.err_exp:
tests/invalid/ho_type_mode_bug.err_exp:
tests/invalid/merge_ground_any.err_exp:
tests/invalid/merge_inst_error.err_exp:
tests/invalid/multimode_dcg.err_exp:
tests/invalid/occurs.err_exp:
tests/invalid/partial_implied_mode.err_exp:
tests/invalid/polymorphic_unification.err_exp:
tests/invalid/qualified_cons_id2.err_exp:
tests/invalid/state_vars_test4.err_exp:
tests/invalid/try_detism.err_exp2:
tests/warnings/unify_f_g.exp:
Update expected outputs.
tests/invalid/*.{m,err_exp}:
tests/misc_tests/*.m:
tests/mmc_make/*.m:
tests/par_conj/*.m:
tests/purity/*.m:
tests/stm/*.m:
tests/string_format/*.m:
tests/structure_reuse/*.m:
tests/submodules/*.m:
tests/tabling/*.m:
tests/term/*.m:
tests/trailing/*.m:
tests/typeclasses/*.m:
tests/valid/*.m:
tests/warnings/*.{m,exp}:
Make these tests use four-space indentation, and ensure that
each module is imported on its own line. (I intend to use the latter
to figure out which subdirectories' tests can be executed in parallel.)
These changes usually move code to different lines. For the tests
that check compiler error messages, expect the new line numbers.
browser/cterm.m:
browser/tree234_cc.m:
Import only one module per line.
tests/hard_coded/boyer.m:
Fix something I missed.
Estimated hours taken: 2
Branches: main
Fix a problem with conjunction re-ordering that was introduced by my
recent change to remove automatic solver type initialisation.
This was causing the stage 2 deep profiler to break.
compiler/modes.m:
Despite its name, modecheck_delayed_solver_goals, in fact
attempts to reschedule all goals, not just solver goals.
Call this predicate as we did before my previous change and
shift the code disables automatic initialisation to a place
where the rescheduling of non-solver goals will not be affected.
tests/valid/Mmakefile:
tests/valid/trace_goal_reorder.m:
A test case for the above bug.
tests/invalid/partial_implied_mode.err_exp:
Revert the last change to this file. The change to the
expected error was caused by the above bug.
Estimated hours taken: 10
Branches: main
Remove support for automatic initialisation of solver types from the language.
This is being done because:
* the current implementation of automatic initialisation means we cannot
support polymorphic solver types, e.g. you cannot have the type foo(bar)
where:
:- solver type foo(T).
:- solver type bar.
* the current initialisation strategy is fairly ad-hoc anyway; in particular
it has a tendency to unnecessarily change the determinism of procedures.
* mode error messages are often quite poor because of the interaction between
automatic initialisation and impure code.
* automatic initialisation is not used in practice. All of the
G12 solver libraries that use solver types recommend explicitly
initialising solver variables.
This change removes support for automatic solver initialisation from the
language. The code for supporting it remains in the implementation, but
it is now dependent upon the developer-only `--solver-type-auto-init'
option.
As a transitional measure the compiler will still accept
`initialisation is ...' attributes in solver type definitions even when
`--no-solver-type-auto-init' is enabled. After this change has bootstrapped,
and the relevant updates have been made to the G12 solver libraries, this
will be changed so that `initialisation is ...' attributes are considered
a syntax error unless `--solver-type-auto-init' is enabled.
doc/reference_manual.texi:
Document that solver type definitions no longer allow initialisation
predicates to be supplied. (The section documenting initialisation
predicates has been commented out rather than deleted since the
implementation of automatic initialisation still exists.)
Remove the section documenting the restrictions on polymorphic
solver types. These restrictions no longer apply in the absence
of automatic initialisation.
compiler/options.m:
Add a new developer-only option, `--solver-type-auto-init', that
controls whether automatic initialisation of solver variables is
allowed (for those solver types that have initialisation predicates
specified.)
compiler/prog_data.m:
Add a type that represents whether a solver type allows automatic
initialisation or not.
Extend the solver_type_details structure to allow initialisation
predicates to be optional.
compiler/prog_io.m:
Allow initialisation predicates to be optional in solver type
definitions.
compiler/modes.m:
compiler/modecheck_call.m:
compiler/modecheck_unify.m:
Only insert calls to solver type initialisation predicates if
the solver type has an initialisation predicate and the
developer-only option `--solver-type-auto-init' is enabled.
compiler/unify_proc.m:
Handle the situation where a solver type does not have an
initialise predicate.
compiler/add_special_pred.m:
Only add initialisation special predicates for those solver
types whose definition provides an initialisation predicate.
compiler/mode_info.m:
Add a utility predicate that tests whether the support for
automatic solver type initialisation is enabled.
compiler/type_util.m:
Add a predicate that tests whether a type is a solver type
that supports automatic initialisation.
Add an XXX comment about such types and abstract equivalence
types.
compiler/mercury_to_mercury.m:
Conform to the above changes.
compiler/special_pred.m:
Fix some typos.
samples/solver_types/eqneq.m:
Delete the `initialisation is ...' from the definition of the solver
type eqneq/1.
tests/hard_coded/Mercury.options:
Enable `--solver-type-auto-init' for the solver_construction_init_test
test.
tests/hard_coded/solver_build_call.m:
tests/invalid/any_pass_as_ground.m:
Don't use automatic solver variable initialisation in these
test cases.
tests/invalid/partial_implied_mode.err_exp:
Conform to the above changes in the mode analyser.
tests/valid/Mercury.options:
Compile some tests cases with `--solver-type-auto-init' enabled.
Estimated hours taken: 2
Branches: main
The following code causes code to be generated which seg-faults
:- interface
:- instance tc(list(T)).
:- implementation.
:- instance tc(list(T)) <= tc(T) where [...].
because the exported instance declaration doesn't contain the
typeclass constraint.
mercury/compiler/add_class.m:
Check that for all the "same" instance declarations
the instance constraints are exactly the same on each
declaration.
tests/invalid/incompatible_instance_constraints*:
Add tests for this code.
tests/invalid/*:
Fix some headvar name errors.
tests/valid/*:
Fix some invalid abstract instance declarations.
Estimated hours taken: 28
Branches: main
A big step towards cleaning up the way we handle errors. The main changes are
- the provision, in error_util.m, of a mechanism for completely specifying
everything to do with a single error in one data structure,
- the conversion of typecheck_errors.m from using io.write_string to
using this new capability,
- the conversion of mode_errors.m and det_report.m from using
write_error_pieces to using this new capability, and
- consistently using the quoting style `symname'/N instead of `symname/N'
in error_util and hlds_error_util (previously, error_util used the former
but hlds_error_util used the latter).
This diff sets up later diffs which will collect all error specifications
in a central place and print them all at once, in order.
compiler/error_util.m:
The new type error_spec, which completely specifies an error.
An error_spec may have multiple components with different contexts
and may have parts which are printed only under certain conditions,
e.g. a given option being set. Each error_spec has a severity
and also records which phase found the error.
The new predicate write_error_spec takes care of updates of the exit
status for errors and (if --halt-at-warn is set) for warnings. It also
takes care of setting the flag that calls for the reminder about -E
at the end.
This diff also makes it simpler to use the ability to print arbitrary
output. It adds the ability to include integers in messages directly,
and the ability to create blank lines. It renames some function symbols
to avoid ambiguities.
Move a predicate that only used by typecheck_errors.m to that file.
compiler/hlds_error_util.m:
Switch to the `symname'/N quoting style for describing predicates and
procedures.
compiler/prog_util.m:
Switch to the `symname'/N quoting style for describing
sym_name_and_arity.
compiler/hlds_module.m:
Provide a predicate to increment the number of errors not by one,
but by the number of errors printed by write_error_spec.
Fix some documentation rot.
compiler/typecheck_errors.m:
Use write_error_spec instead of io.write_strings to print error
messages. In several cases, improve the formatting of the messages
printed.
Mark a number of places where we don't (yet) update the number of
errors in the module_info correctly.
Rename the checkpoint predicate to avoid potential ambiguity with
similar predicates in e.g. mode_info.
compiler/typecheck_info.m:
Group the code for writing stuff out together in one bunch. For each
such predicate, create another that returns a list of format components
instead of doing I/O directly.
compiler/typecheck.m:
Move the code for writing inference messages here from
typecheck_errors.m, since these messages aren't errors.
compiler/mode_errors.m:
compiler/det_report.m:
Use write_error_spec instead of write_error_pieces. In the case of
mode_errors.m, this means we now get correct the set of circumstances
in which we set the flag that calls for the reminder about -E.
compiler/add_pragma.m:
compiler/add_type.m:
Convert some code that used to use write_error_pieces to print error
messages to use write_error_spec instead.
compiler/assertion.m:
compiler/hlds_pred.m:
compiler/post_typecheck.m:
Assertion.m used to contain some code to check for assertions in the
interface that mention predicates that are not exported. Move most
of this code to post_typecheck.m (which is where this code used to be
called from). One small part, which is a test for a particular property
of import_statuses, is moved to hlds_pred.m to be with all the other
similar tests of import_statuses.
compiler/prog_util.m:
Change unqualify_name from a predicate to a function.
compiler/pred_table.m:
compiler/hlds_out.m:
Avoid some ambiguities by adding a suffix to the names of some
predicates.
compiler/*.m:
Conform to the changes above.
library/list.m:
Add a function that was previously present (with different names)
in two compiler modules.
tests/hard_coded/allow_stubs.exp:
Update the format of the expected exception.
tests/invalid/errors2.err_exp2:
Remove this file. As far as I can tell, it was never the correct
expected output on the main branch. (It originated on the alias branch
way back in the mists of time.)
tests/invalid/*.err_exp:
tests/invalid/purity/*.err_exp:
tests/warnings/*.exp:
Update the format of the expected error messages.
tests/recompilation/*.err_exp.2:
Update the format of the expected messages about what was modified.
Estimated hours taken: 10
Branches: main, release
Do not allow discriminated unions with a single zero-arity constructor to have
user-defined equality or comparison. Defining such types causes an assertion
failure in the compiler because the types are considered to be dummy types and
the runtime currently doesn't support (and probably won't ever) d.u. dummy
types with user-defined equality or comparison.
Fix another bug where the compiler was not printing out the `recompile with
-E' prompt at the appropriate time. The bug was caused by the fact that there
were several copies of the globals structure and the one that was being
checked at the time the prompt was being printed out was not the one that had
been updated during the rest of compilation.
compiler/add_types.m:
Emit an error message if an attempt is made to define a d.u. dummy
type with user-defined equality or comparison.
compiler/globals.m:
Remove the extra_error_info field from the globals structure and turn
it into a mutable. Export access predicates for this mutable. The
reason for doing this is that the compiler was sometimes looking at
the wrong copy of the globals structure when checking the value of
this flag - this meant that sometimes the recompile with `-E' prompt
was not being displayed. Turning this flag into a mutable avoids the
problem because now there is only one copy.
compiler/make_hlds.m:
s/__/./ in a few spots.
doc/reference_manual.texi:
Mention the new restrictions on discriminated union types with
user-defined equality or comparison.
tests/invalid/exported_unify2.m:
tests/invalid/exported_unify3.m:
Change some types with user-defined equality or comparison so that
they are no longer dummy types. These test cases have not been
triggering the assertion failure in the compiler because they are only
error checked and the assertion that is failing occurs further along
in the compilation process.
tests/invalid/user_eq_dummy.{m,err_exp}:
Test the new error message for dummy types with user-defined equality
or comparison.
tests/invalid/extra_info_prompt.{m,err_exp}:
Test that the recompile with `-E' prompt is being displayed when it
should.
tests/invalid/Mercury.options:
tests/invalid/Mmakefile:
Include the new test cases.
Where there is a verbose version of the error message compile with
`-E'.
tests/recompilation/add_type_re.err_exp.2:
tests/invalid/*.err_exp:
Update expected outputs to conform to the above.
Estimated hours taken: 1.5
Branches: main
Do not display the `For more information try recompiling with `-E'' prompt
unless we really mean it, i.e. there is actually more information available.
XXX This change is incomplete for the mode_errors module because that
module requires more substantial changes to make this work - I'll do
that as a separate diff.
compiler/globals.m
Add a new global (and access predicates) that keeps track of whether
we have any verbose error information that could be displayed if we
recompiled with `-E'.
compiler/mercury_compile.m
Check the new global flag before prompting the user to recompile with
`-E'.
compiler/mode_errors.m
Add an XXX comment about needing to respect the extra error info flag
properly.
compiler/accumulator.m
compiler/add_clause.m
compiler/add_pred.m
compiler/add_type.m
compiler/assertion.m
compiler/check_typeclass.m
compiler/det_report.m
compiler/magic_util.m
compiler/make_hlds_error.m
compiler/modes.m
compiler/module_qual.m
compiler/modules.m
compiler/post_typecheck.m
compiler/purity.m
compiler/stratify.m
compiler/typecheck_errors.m
Set the new global flag when we come across an error
for which we have a verbose error message.
tests/recompilation/*:
tests/invalid/*:
Update expected error files.
Estimated hours taken: 8
Branches: main
Move the rest of mode_errors.m to use error_util.
compiler/mode_errors.m:
The parts of this module that generate error messages (as opposed to
progress messages) now all return a description of the error to a
central place for printing by error_util.m. This should make it
significantly easier to add new error messages.
compiler/error_util.m:
Add the new capability to support mode_errors.m: that of describing
in one data structure a sequence of calls to write_error_pieces.
compiler/hlds_out.m:
Given a bunch of existing predicates that print various things,
provide versions that convert those things to strings, for use
in mode_errors.m.
Write_unify_context had two versions for printing and only one version
for conversion to pieces; add the second version for conversion to
pieces. Change the order of arguments of the five-argument version
of write_unify_context to allow the use of state variables.
compiler/mercury_to_mercury.m:
compiler/prog_out.m:
Given a bunch of existing predicates that print various things,
provide versions that convert those things to strings, for use
in hlds_out.m and mode_errors.m.
compiler/det_report.m:
Conform to the changed argument order of write_unify_context.
library/term_io.m:
Fix an old bug: the code of add_escaped_char wasn't actually doing
any escaping. hlds_out.m now relies on it doing so.
tests/hard_coded/xmlable_test.m:
Due to the bugfix in term_io.m, string__string now protects &s with
backslashes; expect this.
tests/invalid/*.err_exp:
Expect mode error messages in the new, better format.
Estimated hours taken: 1
Branches: main
compiler/mode_errors.m:
When printing error messages involving lists of insts, put each inst
on a line of its own. Since insts can be deeply nested, it could be
difficult to divide the generated humongous lines into the original
sequence of insts without this diff.
tests/invalid/*.err_exp:
Expect the new format of error messages.
tests/invalid/partial_implied_mode.err_exp:
Update the expected output for this test case as
it has changed since solver types were added.
(Ralph posted a fix for this but didn't commit it).
Estimated hours taken: 0.5
Branches: main
invalid/*.err_exp:
Update the expected output for these test cases since
the compiler now generates error messages that doesn't
use the deprecated syntax for mode declarations.
Estimated hours taken: 14
Branches: main
Change the compiler and tools so that `.' and not `:' is now used as the
module separator in all output.
Infix `.' now has associativity yfx and priority 10.
NEWS:
Report the change.
configure.in:
Amend the test for an up-to-date Mercury compiler to check whether
it recognises `.' as a module qualifier.
compiler/code_gen.m:
compiler/error_util.m:
compiler/hlds_out.m:
compiler/prog_out.m:
compiler/prog_util.m:
compiler/rl_exprn.m:
compiler/rl_gen.m:
compiler/source_file_map.m:
compiler/unused_args.m:
library/io.m:
library/rtti_implementation.m:
library/type_desc.m:
runtime/mercury_debug.c:
runtime/mercury_deconstruct.c:
runtime/mercury_stack_trace.c:
Change `:' to `.' as module separator for output.
compiler/mercury_to_mercury.m:
compiler/prog_io_typeclass.m:
As above.
Fixed a bug where `.' was not being recognised as a module separator.
doc/reference_manual.texi:
Report the change.
library/term_io.m:
Ensure that infix `.' is written without surrounding spaces.
tests/hard_coded/dot_separator.m:
tests/hard_coded/dot_separator.exp:
tests/hard_coded/Mmakefile:
Test case added.
Estimated hours taken: 0.1
tests/invalid/partial_implied_mode.err_exp:
Update now that the head variables of compiler-generated procedures
have real names.
Estimated hours taken: 1
compiler/hlds_goal.m:
Added `set_goal_contexts', which sets all the contexts
of the goal_infos of the sub-goals of a goal.
compiler/unify_proc.m:
Use `set_goal_contexts' to ensure that error messages
for automatically generated procedures have a useful context.
tests/invalid/partial_implied_mode.err_exp:
Update the expected output.
Estimated hours taken: 0.1
tests/invalid/partial_implied_mode.err_exp:
Update the line numbers which changed after Zoltan's
change to stop putting #line directives in .int files.
Estimated hours taken: 10
Fix a code generator abort reported by Stephane Marie
<stephane.marie@detexis.thomson-csf.com>.
The bug was in the code to work out the instmap after a deconstruction
unification with complicated sub-unifications.
The instmap_delta for the unification contained only the bindings for
the first argument for which a new variable was created.
The test case now gives an error during mode analysis
rather than a code generator abort (for unification procedures
of imported types) or a mode error during unique mode analysis
(for local types).
compiler/modes.m:
compiler/modecheck_call.m:
compiler/modecheck_unify.m:
Simplify the code to handle extra goals by not attempting
to bodge together the instmap after the main goal.
Instead, the code now creates the extra unifications and
rechecks the goal with the unifications added.
compiler/mode_info.m:
Add a field to say whether we are reprocessing a goal after
adding extra goals. Abort if the reprocessing adds more extra goals.
The `may_changed_called_proc' field is now separate from
the `how_to_check_goal' field. This is used to avoid
repeating the search for the best procedure when rerunning
mode analysis after adding the extra goals.
compiler/modecheck_unify.m:
Removed predicate unify_vars - it is no longer used.
compiler/unify_proc.m:
Call module_info_remove_predid if there are mode errors
in a unification procedure to avoid spurious determinism
errors.
tests/hard_coded/Mmakefile:
tests/hard_coded/partial_implied_mode.m:
tests/hard_coded/partial_implied_mode.err_exp:
Test case. It would be nicer if the error message pointed
to the unification which caused the unification procedure
to be created, rather than the type declaration in the
interface file.
