Estimated hours taken: 5
Branches: main
compiler/prog_io.m:
More cleanups. Replace Term0, Term1 etc with meaningful variable
names, and use better variable names in general. Rename some
predicates as well when the old names didn't tell you what the
predicate did, and where the name was ambiguous. Convert several
predicates to disjunctions where this allows factoring out common
code or giving consistent names to arguments.
Inline some more auxiliary predicates whose purpose of indexing
on Prolog systems has long expired. In one case, the simplification
that this made possible showed an opportunity to generate more
comprehensive error messages.
Reorder the predicates in this module to form blocks of code each
dealing with one kind of Mercury construct, e.g. type definitions,
mode definitions, predicate and function declarations, etc.
compiler/prog_io_util.m:
A few changes of the same sort as in prog_io.m.
compiler/prog_io_goal.m:
Conform to the new name of a predicate in prog_io.m.
tests/invalid/func_errors.err_exp:
Expect the more comprehensive error messages.
Estimated hours taken: 40
Branches: main
Switch to using error_util to generate error message during the process of
converting terms to prog_items.
In many predicates, we used to return error messages as a string/term pair,
with the string being the error message and a term, which both provided
the context and was printed after the message. We now return error indications
as lists of error_specs. These include a printout of the relevant term only
if this helps users understand the nature or the location of the error.
To make the printouts easier to understand we print variable names in them
using the applicable varsets. (The old version of the compiler used to print
each error term long after it lost track of the right varset, and thus used
a dummy varset that yielded error messages referring to _1, _2 etc instead
of the variable names used by the programmer.)
Sometimes the callers of some parse predicates prepended other strings
indicating the context of the error in front of the error string.
This diff changes things so that now the caller instead passes a list
of format components describing the context to the predicates that construct
the error_specs.
In some places, simplify the code, e.g. by factoring out common code, and by
inlining some auxiliary predicates (we used to need these auxiliary predicates
for indexing when we executed the compiler using Prolog, but those days are
long past).
Mark with XXXs places where I think the error messages or their contexts
could be improved, and places where the structure of the code could be
improved.
compiler/prog_io_util.m:
Change the representation of the maybeN types to use error_spec lists.
compiler/prog_io.m:
compiler/prog_io_dcg.m:
compiler/prog_io_goal.m:
compiler/prog_io_pragma.m:
compiler/prog_io_typeclass.m:
compiler/prog_io_util.m:
Change the way we generate error messages along the lines described
at the top.
In several cases, this required adding extra arguments (varsets,
context descriptions) to predicates for use in error messages.
Some of these predicates were also used in contexts where the caller
was interested only in success, and would ignore any error messages.
In these cases, add a version of the predicate that does not require
the extra arguments, and which is semidet (to allow the caller to
avoid a test for ok).
compiler/error_util.m:
Add a mechanism for changing the case of the next format_component,
to allow an error message to be appended to a list of format_components
providing the context that generates good-looking output whether or not
that context is empty.
Replace some bools with purpose-specific types.
Make sort_error_specs internal to the module, since outside modules
should never need to use it.
Use cords instead of reversed lists to simplify some parts of the
internal implementation.
compiler/mercury_to_mercury.m:
Provide a mechanism to print out terms only if they aren't too big,
for use in our error messages.
compiler/prog_item.m:
Delete the message_list type, and note a future improvement.
compiler/prog_out.m:
Delete the predicates for printing message_lists.
compiler/intermod.m:
compiler/modules.m:
Change the way we print out error messages along the lines described
at the top.
compiler/add_clause.m:
compiler/field_access.m:
compiler/recompilation.check.m:
compiler/recompilation.version.m:
compiler/superhomogeneous.m:
Conform to the changes above by modifying how we generate error
messages.
compiler/add_class.m:
compiler/add_pragma.m:
compiler/check_typeclass.m:
compiler/common.m:
compiler/make.module_dep_file.m:
compiler/make_hlds_error.m:
compiler/make_hlds_passes.m:
compiler/mercury_compile.m:
compiler/mode_errors.m:
compiler/modes.m:
compiler/options_file.m:
compiler/prog_ctgc.m:
compiler/prog_event.m:
compiler/purity.m:
compiler/trans_opt.m:
compiler/typecheck.m:
Trivial updates to conform to the changes above.
compiler/prog_data.m:
Add some field names and access functions for use in the modules above.
library/list.m:
Add list.contains, which is list.member with the arguments reversed
to make it possibly to partially apply it.
tests/invalid/bad_finalise_decl.err_exp:
tests/invalid/bad_initialise_decl.err_exp:
tests/invalid/bad_mutable.err_exp:
tests/invalid/bigtest.err_exp:
tests/invalid/conflicting_fs.err_exp:
tests/invalid/constrained_poly_insts.err_exp:
tests/invalid/errors.err_exp:
tests/invalid/func_errors.err_exp:
tests/invalid/fundeps_unbound_in_ctor.err_exp:
tests/invalid/fundeps_vars.err_exp:
tests/invalid/impl_def_literal_syntax.err_exp:
tests/invalid/inst_list_dup.err_exp:
tests/invalid/invalid_typeclass.err_exp:
tests/invalid/kind.err_exp:
tests/invalid/null_char.err_exp:
tests/invalid/pragma_source_file.err_exp:
tests/invalid/predmode.err_exp:
tests/invalid/reserve_tag.err_exp:
tests/invalid/some.err_exp:
tests/invalid/specified.err_exp:
tests/invalid/trace_goal_env.err_exp:
tests/invalid/type_vars.err_exp:
tests/invalid/typeclass_test_1.err_exp:
tests/invalid/typeclass_test_11.err_exp:
tests/invalid/typeclass_test_2.err_exp:
tests/invalid/unbound_type_vars.err_exp:
tests/invalid/unicode1.err_exp:
tests/invalid/unicode2.err_exp:
tests/invalid/uu_type.err_exp:
tests/invalid/vars_in_wrong_places.err_exp:
tests/invalid/with_type.err_exp:
tests/invalid/purity/purity_nonsense2.err_exp:
Update the expected error messages.
Estimated hours taken: 5
Branches: main
Fix a bug reported by Doug Auclair.
library/ops.m:
Change the Mercury arg priority to 999. It was previously set to
1201, which allowed arguments and list elements to be parsed the
way we intended. However, this had the unintended side effect that
valid terms were written out using incorrect syntax.
library/parser.m:
Place the workaround here, where arguments and list elements are
read.
Note that this is still technically buggy, because we effectively
allow syntax which should be illegal according to the operator
precedence table. But the proper fix for this involves moving to
a BNF style parser, so that is left for later work.
tests/hard_coded/Mmakefile:
Run the term_io_test case twice, checking it against the expected
output each time. The first time it is run with the supplied input
file. The second time it is run with the expected output as input --
this ensures that terms are read in the same way that they are
written out.
tests/hard_coded/term_io_test.m:
Read input from stdin rather than a fixed file.
tests/hard_coded/term_io_test.inp:
Add additional input terms.
tests/hard_coded/term_io_test.exp:
tests/invalid/*.err_exp:
Update the expected output for this and other test cases.
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: 4
Allow user-defined operator precedence tables for parser__read_term
and term_io__write_term.
library/ops.m:
Define a typeclass `op_table' with methods for accessing
operator precedence tables.
Define a type `ops__mercury_op_table', representing the
standard Mercury operator table.
Make `ops__mercury_op_table' an instance of `op_table'.
Define `ops__table' as equivalent to `ops__mercury_op_table'
(it's obsolete, but we can't mark types as obsolete).
Rename `ops__init_op_table' as `ops__init_mercury_op_table',
mark ops__init_op_table as obsolete
library/parser.m:
library/term_io.m:
Add variants of `parser__read_term' and `term_io__write_term'
which take an operator precedence table.
Don't hard-code the maximum priority and argument priority.
Use the `op_table' methods to find those.
Make the priority of operator terms (X `op` Y) 1, not 100.
The reference manual states that operator terms have
the highest precedence possible.
This change slows down a program which does nothing but
parse terms by a bit under 5%, less for writing.
library/hash_table.m:
Fix a few places where parentheses are required because
operator terms now have the lowest possible priority.
`rem' is an operator -- it doesn't need backquotes.
compiler/mercury_to_mercury.m:
library/io.m:
Rename `ops__init_op_table' to `ops__init_mercury_op_table'.
Pass the `op_table' to `ops__max_priority'.
NEWS:
doc/reference_manual.texi:
Document the changes.
Add operator terms to the operator table.
In the "Terms" section of the reference manual, use the same
terminology to describe operator terms as is used in the
"Builtin Operators" section.
samples/Mmakefile:
samples/README:
samples/calculator2.m:
An example program.
tests/hard_coded/term_io_test.exp:
tests/invalid/func_errors.err_exp:
tests/invalid/inst_list_dup.err_exp:
tests/invalid/predmode.err_exp:
tests/invalid/some_err.exp:
`term_io__write_term' now has the same argument priority
behaviour as `parser__read_term', so remove some unnecessary
parentheses from the output.
Estimated hours taken: 12
Allow functions to be declared by supplying the determinism but not the
modes (we assume the default modes).
compiler/make_hlds.m:
Assume default modes if determinism is declared without modes
in the :- func declaration.
compiler/prog_io.m:
Don't give an error message for determinism without modes for
function.
compiler/prog_io_goal.m:
Update documentation to reflect default mode syntax.
Assume default modes if determinism is declared without modes
in lambda expressions.
compiler/prog_util.m:
Mention that declaring modes for some but not all of the
arguments of a function will be noticed in prog_io, so shouldn't
be a concern in this code.
doc/reference_manual.texi:
Document the new syntax.
tests/valid/Mmakefile:
tests/valid/func_default_modes.m:
A test case for the new syntax.
Estimated hours taken: 0.5
tests/invalid/some.err_exp:
tests/invalid/predmode.err_exp:
tests/invalid/func_errors.err_exp:
Update the test results to reflect the extra parentheses
now printed out by term_io__write_term.
Estimated hours taken: 0.25
tests/invalid/Mmake:
tests/invalid/func_errors.m:
tests/invalid/func_errors.err_exp:
New test case to check that we correctly report errors
for certain invalid function declarations.
(Mercury 0.6 failed this test case.)
