Estimated hours taken: 3
Branches: main
Fix a bug introduced by petdr's Dec 2 bug fix which introduced wrapper
functions for unify and compare procs to unbox their arguments,
which caused the tests/valid/unify_typeinfo_bug test case to fail.
The bug was that with --no-special-preds, we were generating invalid
MLDS code, because the code to introduce wrapper functions assumed
that the function being wrapped was defined in the same module,
but with --no-special-preds, the procedure being wrapped is always
private_builtin.unused/0.
In the case of a procedure with arity zero, the wrapper is not needed
(it's just a tail call), so the simplest fix was to just optimize
that case.
compiler/rtti_to_mlds.m:
If the procedure being wrapped has arity zero, don't bother to
generate a wrapper function, since it is not needed.
Estimated hours taken: 0.25
Branches: main
browser/Mercury.options:
trace/Mmakefile:
Ensure that libmer_browser.so and libmer_trace.so link with
the mer_mdbcomp library. This is needed after my recent fix
to ensure that the mer_mdbcomp library is referenced from the
mer_browser library, rather than being included in it.
compiler/Mmakefile:
Add a missing dependency on libmer_mdbcomp.$A.
Estimated hours taken: 0.5
Branches: main
Some Mmakefile tweaks for grade java.
boehm_gc/Mmakefile:
browser/Mmakefile:
trace/Mmakefile:
Don't try to build the boehm_gc, browser, or trace directories
in grade java.
compiler/Mmakefile:
profiler/Mmakefile:
deep_profiler/Mmakefile:
Avoid dependencies on *.$A files in grade java.
Estimated hours taken: 0.5
Branches: main
This just updates some syntax; it doesn't change anything else.
compiler/lp.m:
Replace old-style lambdas with new-style ones.
Estimated hours taken: 2
Branches: main
Fix a bug in the termination analyser where setting the termination
norm to `num-data-elems' causes the an internal abort when analysing
code involving existential typeclass constraints.
The bug is caused by the length of the list of arguments of a
functor differing from the length of the list in the weight table that
tells the compiler which arguments to count when computing the size
of that functor.
The length mismatch is caused by typeinfo related variables that
are introduced by the compiler for existentially typed terms. The
termination analyser includes them but the weight table does not.
I committed a diff a few months ago that partially fixed this problem,
but programs that use existential typeclass constraints break that fix
as well.
The diff implements the easiest solution to all this which is to
have the termination analyser remove all the typeinfo related arguments
of a term before calling term_norm.functor_norm/9.
This diff also fixes a few things in the tests/term directory, namely
making sure that we actually run the tests, updating the module qualifier
in a few .trans_opt_exp files and updating some comments.
compiler/term_norm.m:
Ignore any typeinfo related arguments that a term has when
building the weight table.
compiler/term_traversal.m:
Remove any typeinfo related arguments from the lists of
arguments and modes before computing the size of a term.
tests/term/Mercury.options:
tests/term/existential_error3.m:
tests/term/existential_error3.trans_opt_exp:
Add a regression test for this bug.
tests/term/Mmakefile:
Make sure the $(TESTS) and $(PROGS) variables are fully defined
before Mmake.common is included, otherwise the tests will not run.
Add the new regression test.
tests/term/existential_error1.m:
tests/term/existential_error2.m:
The code that caused these bugs has been moved from term_util.m
to term_norm.m. Update the references to the old filename.
tests/term/*.trans_opt_exp:
Replaces instances of `:' as the module qualifier with `.'.
Quite a few tests cases were failing because the .trans_opt
files use the latter.
Estimated hours taken: 1
Branches: main
Fix a bug where we were incorrectly pruning nondet goals with a non-local
variable of inst "any" whose type contained a solver type in one of
its fields.
compiler/inst_match.m:
Fix a bug: when memoizing the results of previous calls to
inst_matches_*, include the maybe(type) argument in the key
of the memo table, since that argument can effect the result.
tests/valid/Mmakefile:
tests/valid/Mercury.options:
tests/valid/solver_type_bug.m:
Add a regression test for the above-mentioned bug.
Estimated hours taken: 6
Branches: main
Fix a bug where the compiler was treating pseudo_imported external
definitions incorrectly.
compiler/hlds_pred.m:
Change the defintion of external in import_status so that it
records the original import status, not just whether it is
exported or not.
compiler/ml_code_gen.m:
Don't generate any mlds representation for special predicates
whose definition is external and pseudo_imported.
compiler/assertion.m:
compiler/hlds_out.m:
compiler/hlds_pred.m:
compiler/intermod.m:
Adapt to the changes of the definition of status.
tests/hard_coded/Mmakefile:
tests/hard_coded/external_unification_pred.exp:
tests/hard_coded/external_unification_pred.m:
Test case.
Estimated hours taken: 0.25
Branches: main
compiler/mlds_to_il.m:
External references are no longer associated with the MC++
instead they are associated with the C# file, and it is
still hacky!
Estimated hours taken: 2
Branches: main
Don't place predicates whose definition is a foreign_proc into the
.opt file, if the foreign_proc can't be embedded directly in the
generated code as the gain is minimal for the extra complexity
in the generated code.
compiler/intermod.m:
For the il backend, don't export MC++ or C# foreign_procs.
Estimated hours taken: 6
Branches: main
More work on the Java back-end. The standard library now compiles
in grade `java', and hello world (the version using io__write_string,
and linked against the standard library in library/*.m, not the
hand-coded one in java/library/*.java) now works!
compiler/make_hlds.m:
Ignore `pragma type_spec' declarations for the Java back-end.
This works around a problem where javac was failing to compile
some of our generated code due to it overflowing limits on
file name length for the names of the .class files for some
nested classes.
compiler/mlds_to_java.m:
Add some comments. Add myself to the "main authors" list.
library/string.m:
Provide Java definitions of string__first_char and
string__unsafe_index. (These are needed for string.append,
which is used by private_builtin.sorry.)
library/io.m:
Provide Java definitions of io__write_{string,int,char,float}/3.
java/runtime/TypeCtorInfo_Struct.java:
Fix a cut-and-paste error.
java/runtime/TypeInfo_Struct.java:
Improve the implementation of the TypeInfo_Struct(Object)
constructor so that it doesn't throw exceptions during
the initialization of the standard library.
java/runtime/FA_TypeInfo_Struct1.java:
Make this type inherit from TypeInfo_Struct.
Estimated hours taken: 0.5.
Branches: main
compiler/accumulator.m:
Add an XXX comment about this module not handling exceptions
and non-termination correctly.
Estimated hours taken: 8
Branches: main
The comparison and unification function pointers stored in the
type_ctor_info must be pointers to functions where all the arguments
are boxed. This wasn't occuring on the IL backend for types which
are value types.
il_compiler/compiler/rtti_to_mlds.m:
Call ml_gen_closure_wrapper to construct a version of the
comparison and unification functions where the arguments are
boxed. The wrapper function simply unboxes the arguments and
calls the actual special pred.
il_compiler/compiler/ml_closure_gen.m:
Adapt ml_gen_closure_wrapper so that it can generate a wrapper
function for special preds.
tests/hard_coded/Mmakefile:
tests/hard_coded/equality_pred_which_requires_boxing.exp:
tests/hard_coded/equality_pred_which_requires_boxing.m:
Add a test case for this bug.
Estimated hours taken: 100
Branches: main
Make definitions of abstract types available when generating
code for importing modules. This is necessary for the .NET
back-end, and for `:- pragma export' on the C back-end.
compiler/prog_data.m:
compiler/modules.m:
compiler/make.dependencies.m:
compiler/recompilation.version.m:
Handle implementation sections in interface files.
There is a new pseudo-declaration `abstract_imported'
which is applied to items from the implementation
section of an interface file. `abstract_imported'
items may not be used in the error checking passes
for the curent module.
compiler/equiv_type_hlds.m:
compiler/notes/compiler_design.html:
New file.
Go over the HLDS expanding all types fully after
semantic checking has been run.
compiler/mercury_compile.m:
Add the new pass.
Don't write the `.opt' file if there are any errors.
compiler/instmap.m:
Add a predicate instmap_delta_map_foldl to apply
a procedure to all insts in an instmap.
compiler/equiv_type.m:
Export predicates for use by equiv_type_hlds.m
Reorder arguments so state variables and higher-order
programming can be used.
compiler/prog_data.m:
compiler/prog_io_pragma.m:
compiler/make_hlds.m:
compiler/mercury_to_mercury.m:
Handle `:- pragma foreign_type' as a form of type
declaration rather than a pragma.
compiler/hlds_data.m:
compiler/*.m:
Add a field to the type_info_cell_constructor cons_id
to identify the type_ctor, which is needed by
equiv_type_hlds.m.
compiler/module_qual.m:
Donn't allow items from the implementation section of
interface files to match items in the current module.
compiler/*.m:
tests/*/*.m:
Add missing imports which only became apparent with
the bug fixes above.
Remove unnecessary imports which only became apparent with
the bug fixes above.
tests/hard_coded/Mmakefile:
tests/hard_coded/export_test2.{m,exp}:
Test case.
tests/invalid/Mmakefile:
tests/invalid/missing_interface_import2.{m,err_exp}:
Test case.
Estimated hours taken: 2
Branches: main
Bug fixes for the Java back-end.
The standard library now almost compiles in grade java.
compiler/mlds_to_java.m:
Output Java foreign declarations before any other code,
to fix a problem with references to the TYPE_CTOR_REP
constants in library/private_builtin.m.
Fix some bugs where we were not being consistent about mangling
the module name when naming the classes used to simulate taking
procedure addresses. This bug broke Java compilation of library/char.m.
Fix some bugs where the compiler was getting confused about
which types map to array types in Java.
For MLDS casts that cast to type_info or pseudo_type_info,
generate Java constructor calls, not Java casts.
This fixes type errors in the generated Java code.
Simplify the code for hand_defined_type.
compiler/rtti.m:
Fix a bug in tc_rtti_name_java_type: map typeclass_infos to
the Java type "java.lang.Object[]", not "java.lang.Integer[]".
The latter didn't work because the elements which hold the method
addresses do not have type java.lang.Integer.
java/runtime/DuExistInfo.java:
java/runtime/NotagFunctorDesc.java:
Define constructors for these types.
java/runtime/TypeInfo_Struct.java:
Define some additional constructors for this type.
library/builtin.m:
Provide Java stub definitions of
- classes for the types func/0 and c_pointer/0;
- unify/compare preds for func, c_pointer, tuple, and void;
Define Java definitions for the additional modes of compare/3
(they just call the first mode).
library/exception.m:
Define Java versions of make_io_state and consume_io_state,
and Java stubs for throw_impl and catch_impl.
Change try_all so that it calls catch_impl, like try does,
rather than calling builtin_catch directly. This is needed
since the .NET and Java back-ends only define catch_impl,
not builtin_catch.
library/io.m:
Provide Java `pragma foreign_type' declarations for the types
array.array(T) and io.system_error. This is needed to avoid
compilation errors when building in grade `java'.
library/private_builtin.m:
Delete the Java definition of the type_info for type_info/1,
because that is generated automatically now,
after petdr's recent bug fix.
Provide Java stubs definitions of the unify/compare preds
for ref/1, heap_pointer/0, type_ctor_info, type_info,
base_typeclass_info, and typeclass_info.
Provide Java definition of dummy_var.
library/type_desc.m:
Provide Java stub definitions of
- classes for type_desc/0 and type_ctor_desc/0
- unify/compare preds for those types
library/store.m:
Define unification and comparison predicates for the store(S)
type. This is needed because the automatically generated ones
might not have the right semantics, but also because the Java
back-end generates some invalid code for the automatically
generated unification and compare predicates (the generated
code has some invalid references to "dummy_var").
Estimated hours taken: 12
Branches: main
Bug fixes for the Java back-end.
compiler/mlds_to_java.m:
Fix a bug where we were outputting an incorrect type for nested
structure initializers
Fix a bug where we were outputting invalid syntax for some
array initializers.
Fix a bug where the code output for string comparisons was not properly
parenthesized.
compiler/mlds.m:
Add a new argument to init_struct that specifies the type of
the structure being initialized. This is needed to handled
nested structure initializers for the Java back-end.
compiler/rtti.m:
Add new alternatives to ctor_rtti_name for all the types which
are used as nested components of RTTI structures. This is
needed in order to have appropriate RTTI types to use in
rtti_to_mlds.m for the new field of init_struct.
compiler/rtti_to_mlds.m:
Substantial changes to fill in the new field of init_struct
in all the generated initializers.
compiler/ml_elim_nested.m:
compiler/ml_optimize.m:
compiler/ml_unify_gen.m:
compiler/ml_util.m:
compiler/mlds_to_c.m:
compiler/mlds_to_gcc.m:
compiler/mlds_to_il.m:
Minor changes to handle the new argument to init_struct.
compiler/mlds_to_gcc.m:
compiler/opt_debug.m:
Minor changes to handle the new alternatives for ctor_rtti_name.
compiler/mlds_to_gcc.m:
Fix a bug where it was generating a structure type, rather
than an array thereof, for res_name_ordered_table.
Estimated hours taken: 1
Branches: main
compiler/polymorphism.m:
Split a predicate with a 200-line definition into three.
Fix some formatting issues.
There is no change in algorithms.
Estimated hours taken: 2
Branches: main
Fix a bug where the type_ctor_info for type_info/1 and
type_ctor_info/1 weren't being generated.
compiler/type_ctor_info.m:
Also have to generate a type_ctor_info for those types which
have hand defined rtti due to having a fake type body.
compiler/type_util.m:
Rename builtin_type_ctors_with_no_type_defn as
builtin_type_ctors_with_no_hlds_type_defn as the former name
was misleading, and remove an incorrect comment about which
type_ctors this function returns.
compiler/make_hlds.m:
compiler/typecheck.m:
Changes due to the renaming of
builtin_type_ctors_with_no_hlds_type_defn.
Estimated hours taken: 1
Branches: main
A bug fix for the Java back-end.
compiler/ml_type_gen.m:
Fix a bug where we were generating invalid Java code for constructors.
For a constructor Foo with an argument of type Bar, we were generating
public Foo(Bar F1) {
((modulename.Foo)this).F1 =
modulename.Foo.F1; // XXX
}
instead of
public Foo(Bar F1) {
((modulename.typename)this).F1 = F1;
}
The fix was to change ml_type_gen.m so that for the Java back-end,
it generates MLDS corresponding to
public Foo(Bar F1) {
((modulename.typename)this).F1 = modulename.F1;
}
and then the "modulename." qualifier on F1 gets removed by the
existing code in mlds_to_java.m.
Estimated hours taken: 12
Branches: main
compiler/handle_options.m:
Disable the dup_elim optimization when accurate GC is enabled,
because it does not correctly preserve the accurate GC layout
information.
Estimated hours taken: 0.5
Branches: main
compiler/constraint.m:
compiler/deforest.m:
compiler/goal_util.m:
compiler/simplify.m:
compiler/rl.m:
s/goal_cannot_loop/goal_cannot_loop_or_throw/ throughout
the optimization passes. This is needed to avoid optimizing
away calls to error/1 now that it is marked as terminating.
compiler/pd_util.m:
Remove a duplicate of goal_util__reordering_maintains_termination.
Estimated hours taken: 1.5
Branches: main
Address review comments of fjh about my change to generate type_ctor and
special preds for the builtin types which don't have an abstract type
declaration.
compiler/make_hlds.m:
Change parse_tree_to_hlds to use the state variable syntax.
compiler/type_util.m:
s/builtin_type_ctors/builtin_type_ctors_with_no_type_defn/
which better reflects which type ctors this function returns.
compiler/type_ctor_info.m:
Improve the documentation of how the code determines whether
or not a type_ctor_info needs to be generated.
compiler/typecheck.m:
Use builtin_type_ctors_with_no_type_defn and make the documentation
more explicit.
Estimated hours taken: 4
Branches: main
Rather than declare the builtin types as abstract types in builtin.m
special case their handling in the compiler to ensure that the
special preds and the type_ctor_infos are generated when needed.
compiler/type_util.m:
Add a utility predicate which returns the list of type_ctors
which represent types which are special builtins (ie they have no type
declaration, even an abstract one).
compiler/make_hlds.m:
Add the special preds for the special builtins if we are generating
the rtti for the builtins.
compiler/type_ctor_info.m:
Add type_ctor_infos for the special builtins if we are generating
the rtti for the builtins.
compiler/typecheck.m:
The special predicates for the special builtins don't require
typechecking.
library/builtin.m:
Remove the special builtins as they are no longer needed for
generating the type_ctor_infos for a type.
Estimated hours taken: 36
Branches: main
Get the compiler to generate the type_ctor_infos for all the types
which were previously defined by hand.
This change completes the removal of all the MC++ from the compiler.
compiler/make_hlds.m:
Remove the builtin qualifier from the builtin types, before
processing them.
compiler/mlds_to_il.m:
Remove the special code which handled those type_ctor_infos
which where defined in MC++.
compiler/special_pred.m:
Add a routine compiler_generated_rtti_for_the_builtins which
returns if the compiler generates RTTI for builtin types, or
they are hand-coded. It returns true for the non-C backends.
Change can_generate_special_pred_clauses_for_type so that it
returns true for the builtin types whose body is defined in
Mercury in private_builtin.m.
compiler/type_ctor_info.m:
Generate a type_ctor_info for the builtin types if needed.
compiler/unify_proc.m:
Generate calls to unification and comparison predicates defined
in private_builtin.m.
library/Mmakefile:
library/builtin.m:
library/private_builtin.m:
library/type_desc.m:
Remove the final MC++ code and rename some of the code to use
the name which the Mercury compiler generates.
Estimated hours taken: 2
Branches: main
compiler/mlds_to_il.m:
Only the RTTI is written in MC++, so any of the variables from
private_builtin are defined in C#.
runtime/mercury_il.il:
Refer to the C# file not the MC++ file.
Estimated hours taken: 3
Branches: main
Convert the last of the code in the library from MC++ to C#.
However the RTTI datastructures for the builtin types are still defined
in MC++, for the moment.
compiler/mlds_to_il.m:
Change the mangling so that for the names that need mangling
that aren't RTTI types we look for them in the C# module.
library/builtin.m:
library/private_builtin.m:
library/type_desc.m:
Convert all the code from MC++ to C#, and mark that the
MR_DEFINE_BUILTIN_TYPE_CTOR_INFO macros need to be rewritten in
Mercury.
Estimated hours taken: 1
Branches: main
Change the calling convention for higher order calls, to make such calls
faster.
The change to mercury_ho_call.c will be committed first. The change to the
compiler will be committed later, after people had a chance to do a "cvs
update" of the change of the runtime. This will avoid installed compilers
generating calls to labels that do not exist in the runtime directories of
workspaces.
runtime/mercury_ho_call.c:
Add versions of mercury_do_call_closure/mercury_do_call_class_method
that do not expect the argument giving the number of output arguments,
since the code does not and will not need this information.
compiler/call_gen.m:
Do not generate the argument giving the number of output arguments,
and start placing user-specified arguments into the freed-up register.
compiler/llds_out.m:
Generate code that calls the compact versions of the labels in
mercury_ho_call.c.
Estimated hours taken: 1.5
Branches: main
Rename mercury_mcpp.cs to mercury_dotnet.cs and switch it to under
control of configure. Switching to under the control of configure
allows us to define some constants determined at configure time.
runtime/mercury_mcpp.cs:
runtime/mercury_dotnet.cs.in:
Copy mercury_mcpp.cs to mercury_dotnet.cs.in and define two
new constants MR_VERSION and MR_FULLARCH which are needed by
library.version/1.
library/library.m:
Rewrite library.version/1 in C#.
configure.in:
Add runtime/mercury_dotnet.cs to the list of files under
control.
compiler/mlds_to_managed.m:
compiler/modules.m:
library/Mmakefile:
library/private_builtin.m:
runtime/Mmakefile:
Updates to reflect the name change from mercury_mcpp to
mercury_dotnet.
Estimated hours taken: 0.25
Branches: main
compiler/notes/coding_standards.html:
w3/information/developers/c_coding_standard.html:
Specify that code should generally be ordered top-down, not bottom-up.
Estimated hours taken: 2
Branches: main
Fix a bug that broke bootstrapping in grade hlc.agc.
compiler/ml_closure_gen.m:
Ensure that we generatine GC tracing code for the closure_arg
parameter. This is needed because it gets referenced by the
GC tracing code for other local variables in this function.
Without this, if two GCs occur between when a closure is
called and when it returns, then the first GC will collect
the closure, and then the second GC will try to dereference
the closure_arg parameter.
Estimated hours taken: 1
Branches: main
compiler/code_info.m:
compiler/var_locn.m:
Add some comments, to address Zoltan's review comments on my
previous change.
Estimated hours taken: 0.1
Branches: main
compiler/mlds_to_managed.m:
Define MR_HIGHLEVEL_DATA, when necessary, at the beginning of
the C# file as we do for MC++.
Estimated hours taken: 0.25
Branches: main
compiler/ml_util.m:
Reorder some code, so that it goes next to the comment which
describes it. Also, fix a misleading comment.
Estimated hours taken: 4
Branches: main
Fix a bug with accurate GC. The mapping from code addresses to stack
layouts was not working properly, because GCC was reordering code within
C functions (LLDS modules) so that a label at end of the module did not
necessarily have the highest address. The solution is to put the end label
in a different C function, i.e. a new LLDS module.
runtime/mercury_goto.h:
Back out previous attempt to solve this problem, since it didn't work.
compiler/transform_llds.m:
If accurate GC is enabled, append a module containing a dummy end label
at the end of the module list.
Estimated hours taken: 0.1
Branches: main
The type private_builtin.ref/1 is abstract, this leads to problems on strongly
typed backends such as the il backend which can generate references to this
type.
compiler/ml_util.m:
Add private_builtin.ref/1 to the list of types which need a lowlevel
representation.
Estimated hours taken: 4
Branches: main
Improve the support for debugging in the il grade, by not optimizing
the generated il code, unless needed for verifiability.
Using
mmc --make --grade il --debug diff
now generates code which is relatively easier to debug in an IL
debugger.
compiler/handle_options.m:
--debug imples --target-debug for the il backend.
if debugging and --no-trace-optimize and il backend then
--no-optimize-peep.
Allow --debug as valid option for the il backend.
compiler/mlds_to_ilasm.m:
Use --verifiable-code and --optimize-peep to decide what level
of peephole optimization is needed to be done.
compiler/il_peephole.m:
Change the peephole optimizer so that it can do all peephole
optimizations, or only those which are needed for
verifiability.
Branches: main
Estimated hours taken: 0.5
Fix a bug that broke the MLDS back-end in the case when either heap profiling
or accurate GC was enabled.
compiler/mlds_to_c.m:
Fix a bug in Zoltan's recent change to declare a local variable
"base" for initializing terms allocated on the heap:
change the order in which code is output so that the
declaration of "base" comes immediately after the "{",
without any intervening code. Otherwise, it is not legal C89.
Estimated hours taken: 6
Branches: main
This diff changes the LLDS backend to make it easier to read and to maintain,
but contains no changes in algorithms whatsoever.
compiler/basic_block.m:
compiler/dupelim.m:
compiler/frameopt.m:
compiler/jumpopt.m:
compiler/layout_out.m:
compiler/livemap.m:
compiler/peephole.m:
compiler/reassign.m:
compiler/rtti_out.m:
compiler/use_local_vars.m:
Convert these modules to our current coding standards. Use state
variable notation when appropriate, reordering arguments as necessary.
compiler/llds_out.m:
Convert these modules to our current coding standards. Use state
variable notation when appropriate, reordering arguments as necessary.
Delete predicates which are just specialized forms of foldl, using
foldl (or foldl2 etc) directly instead.
Factor out some common code.
compiler/livemap.m:
Convert these modules to our current coding standards. Use state
variable notation when appropriate, reordering arguments as necessary.
Remove some special case handling that used to be required by the value
numbering pass.
library/bintree_set.m:
Provide a function version of the initialization predicate.
Estimated hours taken: 1
Branches: main
This diff changes goal_util.m to make it easier to read and to maintain,
but contains no changes in algorithms whatsoever.
compiler/goal_util.m:
Convert this module to our current coding standards. Use state
variable notation when appropriate, reordering arguments as necessary.
compiler/*.m:
Conform to the changes in the above module.
Estimated hours taken: 1
Branches: main
compiler/mlds_to_c.m:
When filling in the fields of a newly allocated cell, use a fresh local
variable as the base address for the field references in preference to
lvals that are more expensive to access. This yields a speedup of about
0.3%. (The LLDS backend has had a similar optimization for a long
time.)
Estimated hours taken: 3
Branches: main
Another step towards supporting accurate GC for the LLDs back-end:
modify the LLDS code generator to allocate the proper forwarding pointer
slots for type_infos and typeclass_infos.
compiler/unify_gen.m:
compiler/code_info.m:
compiler/var_locn.m:
When allocating type_infos or typeclass_infos on the heap,
if accurate garbage collection is enabled then ensure that we
reserve a word before the start of the object. This word is
used by the collector to store a forwarding pointer.
For ordinary objects, we store the forwarding pointer in the
old copy of the object, but we can't clobber the old copies of
type_infos and typeclass_infos, since they might be used _during_
garbage collection.
Estimated hours taken: 4
Branches: main
compiler/make_hlds.m:
Fix a bug where the state variable transformation was not
correctly handling the case where the condition of an if-then-else
referred to a state variable, but the "then" part didn't.
(Also, fix some incorrect comments for get_rev_conj and get_rev_disj,
and rearrange some misleading whitespace in the comments for the
svar_info type.)
tests/valid/Mmakefile:
tests/valid/state_var_bug.m:
Add a regression test.
Estimated hours taken: 6
Branches: main
This diff changes modes.m, unique_modes.m and mode_info.m to make them easier
to read and to maintain, but contains no changes in algorithms whatsoever.
compiler/modes.m:
compiler/unique_modes.m:
compiler/mode_info.m:
Convert these modules to our current coding standards. Use state
variable notation when appropriate, reordering arguments as necessary.
Delete predicates whose only purpose was to ease programming with DCGs.
Remove the IO state from the mode_info structure, and pass it
separately to the (relatively few) predicates that need it. This
avoids using complicated (and as yet unsupported) modes or lying
to the compiler (we used to do the latter), and we are no longer
forced into that choice by being limited to a single hidden (DCG)
variable.
Use more expressive variable names and factor out code as appropriate.
compiler/*.m:
Conform to the changes in the above two modules.
Estimated hours taken: 4
Branches: main
This diff contains no changes in algorithms whatsoever.
browser/*.m:
compiler/*.m:
library/*.m:
Replace old-style lambdas with new-style lambdas or with named
procedures.
