compiler/module_qual.qual_errors.m:
When generating the context for an error message of the form
"In definition of type/inst/mode/pred/func X:", if the X is guaranteed
to be defined in the current module, then print its name unqualified.
Since the module qualifier is guaranteed to be the same as the name
of the module being compiled, its presence in the error message
adds no information and is therefore only clutter.
tests/invalid/errors.err_exp:
tests/invalid/errors1.err_exp:
tests/invalid/int_impl_imports.err_exp:
tests/invalid/kind.err_exp:
tests/invalid/test_nested.err_exp:
tests/invalid/type_arity.err_exp:
tests/invalid/undef_type.err_exp:
tests/invalid_make_int/missing_interface_import.int_err_exp:
tests/invalid_make_int/missing_interface_import.int_err_exp2:
Expect the updated error message.
tests/invalid_make_int/missing_interface_import.m:
Fix a comment.
NEWS:
Announce this fact.
compiler/modules.m:
If module m.n imports module x.y.z, require it to import its ancestor
modules (x and x.y).
If module m.n imports module x.y.z in the interface, require it to import
its ancestor modules (x and x.y) in the interface.
If module m.n imports module x.y.z, require us to see an include for z in
x.y, if we see an interface file for x.y.
Given a module m.n, require us to see an include for n in m's private
interface file.
These rules are tighter than the rules we enforced before.
Report any violations of the above rules when generating code for m.n.
Previously, we also reported violations when generating interface files.
compiler/prog_item.m:
For each item block, record which module it came from. (We have long
recorded which *section* of that module it came from.) This is needed
by some of the new functionality in modules.m.
Likewise, record the module name in the module components (from which
item blocks are later constructed).
compiler/parse_module.m:
Record the module name in module components and item blocks.
compiler/check_raw_comp_unit.m:
compiler/comp_unit_interface.m:
compiler/deps_map.m:
compiler/equiv_type.m:
compiler/get_dependencies.m:
compiler/make.module_dep_file.m:
compiler/make_hlds_separate_items.m:
compiler/module_imports.m:
compiler/module_qual.collect_mq_info.m:
compiler/module_qual.m:
compiler/module_qual.qualify_items.m:
compiler/parse_tree_out.m:
compiler/prog_item_stats.m:
compiler/recompilation.check.m:
compiler/split_parse_tree_src.m:
Conform to the changes above.
tests/invalid/missing_parent_import.err_exp:
Expect an additional error message we now generate.
tests/invalid/sub_c.err_exp:
tests/invalid/test_nested.err_exp:
Expect the more specific error messages we now generate.
If a module m1 imports module m2 using a use_module declaration in its
interface section but also imports it using an import_module declaration
in its implementation section, then references to entities defined in m2
in the interface of m1 must be module qualified, but similar references
in the implementation of m1 need NOT be module qualified.
The Mercury compiler has long had a bug that allowed entities imported
in the implementation of module m1 to be used in the interface of m1,
against the rules of the language. When I fixed that bug on 2015 nov 11,
the compiler became unable to properly process the situation described
in the paragraph above, because it still had another bug, which was that
it had a single setting for module qualification requirements: either it
was required *everywhere* in m1, or it was required *nowhere* in m1.
This diff fixes that bug (mantis bug 401) by allowing different requirements
in this respect in the interface of m1 versus its implementation.
It also changes the de-facto meaning of what exactly a use_module declaration
requires. The reference manual says only:
Uses of entities imported using @code{use_module} declarations
@emph{must} be explicitly module qualified.
This WAS unambiguous when it was written in june 1997, a few months before
submodules were added to the language. With a flat module system, there were
no nested module names, so either the module name was present, or it wasn't.
In the presence of nested modules, though, it IS ambiguous: HOW MUCH of
the module name must be present? The reference manual is silent on this.
At the moment, in the presence of a use_module for module ma.mb.mc,
a predicate p1 in ma.mb.mc may of course be referred to as ma.mb.mc.p1,
but may also be referred to as mb.mc.p1, or even just mc.p1. The specification
of mc is always required, but exactly WHICH of the previous module name
components are required depends on whether e.g. ma or ma.mb were themselves
imported via use_module or import_module declarations. The code implementing
the check (find_matches_in_permissions_map in module_qual.id_set.m) is itself
not too clear.
This diff changes the rules so that all references to anything imported
via a use_module declarations have to be fully qualified. This is stricter
than the existing rule, but it is MUCH simpler, and it better matches
the original intent of use_module declarations, which is the avoidance
of ALL POSSIBILITY of ambiguity.
NEWS:
doc/reference_manual.texi:
Document this change to the language.
compiler/prog_item.m:
Change the representation of sections in .int* files. Previously,
we recorded whether the interface file that the section was from
was read for an `import_module' or a `use_module' declaration.
We now allow the recording of the fact that it had both declarations,
a use_module in the interface and an import_module in the implementation.
compiler/modules.m:
Treat modules that have both a `use_module' declaration in interface
and an `import_module' declaration in the implementation as being of this
new internal section kind.
Add some infrastructure for debugging similar problems in the future.
compiler/module_qual.id_set.m:
Change the permissions system. Instead of recording *independently*
whether (a) an entity may be used in the interface and (b) whether
references to it must be qualified, record *separately* whether
(a) it may be used in the interface, and if so with what qualification
requirements, and (b) what its qualification requirements are in the
implementation. Unlike the old permission setup, the new one allows us
to express the "use_module in interface, import_module in implementation"
situation.
Implement the change to the language, by requiring full module
qualification of entities imported via use_module declarations.
This allows us to use a single parameterized piece of code to handle
both unqualified and qualified references, whereas before we used
separate pieces of code for them.
Change the way we handle errors in module qualification. Instead of
looking only for precise matches only, and looking for the possible causes
of errors only if we find either no matches or two or more matches, we now
return the near-miss matches as well, the ones that we need to decide
on what error message we want to generate. While this is slightly slower,
it is *much* simpler, since it allows us to avoid duplicating the search
code: once for precise matches only, once for no-match errors, and once
for multiple-match errors.
compiler/module_qual.collect_mq_info.m:
Set permissions for symbol access according to the new permissions system.
Code that previously triggered the bug will use the new interface file
section kind, and can now have its permissions record correctly.
compiler/module_qual.qual_errors.m:
Use a single predicate to generate error messages for all situations
in which there is not a single precise match for a module qualification.
This allows us to diagnose multiple potential problems with a single
lookup.
Change the wording of an existing error message to avoid a possibly-untrue
implication.
compiler/hlds_data.m:
Rename a field of the type_defn type to better document its meaning.
compiler/make_hlds.m:
Document the meaning of the need_qualifier field in sec_infos.
compiler/add_type.m:
compiler/make_hlds_separate_items.m:
compiler/module_qual.m:
compiler/pred_table.m:
compiler/parse_tree_out.m:
compiler/prog_item_stats.m:
Conform to the above changes.
mdbcomp/sym_name.m:
Add a utility predicate for use in looking for near-miss matches
in module qualification.
Make the documentation of a related predicate more precise.
tests/invalid/bad_instance.m:
tests/submodules/class.m:
tests/submodules/nested.m:
tests/submodules/nested3.m:
tests/submodules/parent.m:
Make names fully module qualified where the language rules now require it.
tests/submodules/deeply_nested.m:
Comment out a part of the test that tested the partially qualified names,
since these are now not allowed.
tests/invalid/errors.err_exp:
Expect a now-improved error message.
tests/invalid/test_nested.err_exp:
tests/invalid/transitive_import.err_exp:
tests/invalid/undef_type.err_exp:
Expect a now less-likely-to-mislead error message.
tests/valid_seq/int_impl_imports.m:
tests/valid_seq/int_impl_imports_2.m:
The test case for mantis bug 401.
tests/valid_seq/Mmakefile:
Enable the new test case.
This fixes mantis bug 363.
compiler/prog_io_pragma.m:
Remove the old code that deliberately used an old-style argument number
(derived from the syntax of the long-ago-removed pragma_c_code pragma,
which did not have a foreign_lang argument) for error messages.
This was mantis bug 363.
Put the code for parsing the main components of foreign_procs in
one place, and parse them in order. If more than one component has
errors, return all their error messages.
Change the code that parses foreign_proc attribute lists to return
an error message for each unrecognized attribute, instead of just
printing "invalid attribute list".
compiler/error_util.m:
Have nth_fixed generate words ("fourth", "fifth" etc) up to N = 10,
and generate e.g. 42nd instead of 42th.
tests/invalid/bug363.{m,err_exp}:
Add an expanded version of the mantis 363 test case.
tests/invalid/Mmakefile:
Enable the new test case.
tests/invalid/test_nested.err_exp:
Conform to the change in error_util.m.
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.
Specifically, when we find undefined types in type definitions, say WHERE
the undefined type is (both as line number and as function symbol/arg number,
and field name if present), since the body of the type definition is sometimes
quite big.
compiler/module_qual.m:
When module qualification found an error, it used to get the context
it printed for the error from the mq_info structure, which also contained
the rest of the state of the qualification process. The drawback of this
setup is that the mq_info's record of the error context was updated
only in a few of the places where that context actually changed.
This diff takes the error context out of the mq_info, and passes it
as a separate argument to the predicates that need it. This makes it
really visible if a context is passed onward unchanged even when it
should be updated.
Fix some of these places, and mark the rest with XXXs.
When printing error messages about predicate or function declarations,
the message talked about definitions, not declarations. Fix that.
Put a prog_context inside each function symbol of mq_error_context;
don't require the creation of a separate memory cell for a pair
to link the mq_error_context with the prog_context.
Factor out some common code.
compiler/prog_data.m:
I tried to put a context inside constraints for use in error contexts
in module_qual.m, but that turned out to be a bad idea. Document why.
compiler/add_class.m:
compiler/equiv_type.m:
compiler/goal_util.m:
compiler/higher_order.m:
compiler/hlds_data.m:
compiler/post_typecheck.m:
compiler/pred_table.m:
compiler/prog_io_typeclass.m:
compiler/prog_type.m:
compiler/recompilation.usage.m:
compiler/type_ctor_info.m:
compiler/type_util.m:
compiler/typecheck.m:
compiler/typeclasses.m:
compiler/unused_imports.m:
compiler/write_module_interface_files.m:
compiler/xml_documentation.m:
Clean up some code dealing with constraints. I did this cleanup while
adding contexts to constraints, a change I then had to undo.
tests/invalid/builtin_int.err_exp:
tests/invalid/errors.err_exp:
tests/invalid/errors1.err_exp:
tests/invalid/fundeps_vars.err_exp:
tests/invalid/missing_interface_import.err_exp:
tests/invalid/missing_interface_import2.err_exp:
tests/invalid/test_nested.err_exp:
tests/invalid/transitive_import.err_exp:
tests/invalid/undef_type.err_exp:
tests/invalid/undef_type_mod_qual.err_exp:
tests/recompilation/add_type_re.err_exp.2:
tests/recompilation/remove_type_re.err_exp.2:
Update these expected output files to match the better messages
we now generate.
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: 6
Branches: main
compiler/module_qual.m:
Convert this module to our current coding style. Use state variables
where appropriate. Reorder arguments as required to allow the use of
state variables. Use error_util instead of io__write_string to print
error messages. Use predmode declarations where possible. Use . as
the module name separator.
There are no algorithmic changes except for the use of error_util.
compiler/middle_rec.m:
Convert this module to our current coding style. Use predmode
declarations where possible.
compiler/make_hlds.m:
compiler/mercury_compile.m:
Conform to changed argument orders of predicates in module_qual.m.
compiler/error_util.m:
Add a new version of the utility predicate that puts commas and "and"s
between elements of a list.
Fix an old bug: when printing a symname/arity pair, put quotes around
only the symname part.
tests/invalid/*.err_exp:
tests/invalid/purity/*.err_exp:
tests/recompilation/*.err_exp.2:
tests/warning/*.exp:
Update the expected error messages to reflect the better wrapping we
get from error_util, and in some cases to reflect the fixed code for
quotes around symnames.
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: 0.25
tests/invalid/missing_interface_import.err_exp
tests/invalid/test_nested.err_exp
Update the error messages for undefined types in a few test cases
to reflect my recent improvemnt to that error message.
Estimated hours taken: 4
Fix a limitation of the current nested module support:
ensure that we check for attempts to import inaccessible modules.
compiler/modules.m:
Check for attempts to import inaccessible modules
(modules whose parent has not been imported,
or for which there is no `include_module'
declaration in the parent's interface).
compiler/intermod.m:
Modify the way we import `.opt' files. The previous method
temporarily set the items field of the module_imports
structure to [], but check_module_accessibility relies
on the assumption that the items field contains the
items for all modules previously read in.
tests/invalid/Mmakefile:
tests/invalid/test_nested.m:
tests/invalid/test_nested.err_exp:
tests/invalid/parent.m:
tests/invalid/parent.private_child.m:
tests/invalid/parent.public_child.m:
tests/invalid/parent.undeclared_child.m:
tests/invalid/parent.undeclared_child.err_exp:
tests/invalid/parent2.m:
tests/invalid/parent2.child.m:
Add some tests for the above change.
doc/reference_manual.texi:
Update the "bugs and limitations" sub-section of the modules
chapter to reflect the new status quo.
