[m-dev.] for review: type specialization [1]

[Fergus Henderson]

On 17-Feb-1999, Simon Taylor <stayl at cs.mu.OZ.AU> wrote:
> Index: compiler/make_hlds.m
> +add_pragma_type_spec(Pragma, SymName, SpecName, Arity, MaybePredOrFunc,
> +		MaybeModes, SpecSubst, VarSet, Context, Module0, Module) -->
> +	{ module_info_get_predicate_table(Module0, Preds) },
> +	(
> +		{ MaybePredOrFunc = yes(PredOrFunc) ->
> +			predicate_table_search_pf_sym_arity(Preds,
> +				PredOrFunc, SymName, Arity, PredIds)
> +		;
> +			predicate_table_search_sym_arity(Preds,
> +				SymName, Arity, PredIds)
> +		}
> +	->
> +		( { PredIds = [PredId] } ->
> +			add_pragma_type_spec_2(Pragma, SymName, SpecName,
> +				Arity, SpecSubst, MaybeModes, VarSet, Context,
> +				PredId, Module0, Module)
> +		;
> +			% XXX we should allow the programmer to specify
> +			% predicate or function to avoid this problem.
> +			% It's difficult to just specialize all matching names
> +			% because we've only included a single specialized
> +			% name in the interface file.
> +			{ Module  = Module0 },
> +			io__set_exit_status(1),
> +			prog_out__write_context(Context),
> +			io__write_string(
> +			    "Error: `pragma type_spec' declaration matches\n"),
> +			prog_out__write_context(Context),
> +			io__write_string(
> +			    "  multiple predicates or functions.\n")
> +		)

That error condition should be documented in the language
reference manual.

> +	;
> +		undefined_pred_or_func_error(SymName, Arity, Context,
> +			"pragma type_spec declaration"),

That should perhaps be "`pragma type_spec' declaration".
Please make sure that it is consistent with whatever format
we use for similar messages for other pragma declarations.

...
> +		% Even if we aren't doing type specialization, we need
> +		% to create the interface procedures for local predicates
> +		% to check the type-class correctness of the requested
> +		% specializations.

The type-class correctness condition referred to here
should be documented in the language reference manual.

> +report_pragma_type_spec(PredInfo0, Context) -->
> +	{ pred_info_module(PredInfo0, Module) },
> +	{ pred_info_name(PredInfo0, Name) },
> +	{ pred_info_arity(PredInfo0, Arity) },
> +	{ pred_info_get_is_pred_or_func(PredInfo0, PredOrFunc) },
> +	prog_out__write_context(Context),
> +	io__write_string("In `pragma type_spec(...)' declaration for "),

Perhaps s/(...)//

Actually I'm not particularly dogmatic about whether the
`(...)' should be there, but please make sure it is
consistent with the messages for other pragmas.

> +handle_pragma_type_spec_modes(SymName, Arity, Context, MaybeModes, ProcIds,
> +		Procs0, Procs, ModesOk, ModuleInfo0, ModuleInfo) -->

A brief comment explaining what that predicate does would be helpful.

...
> +			undefined_mode_error(SymName, Arity, Context,
> +				"`:- pragma type_spec(...)' declaration"),

Again, please make sure that the use of `:-' and `(...)' here
are consistent with what is done for other pragmas.

Apart from those things, the rest of part [1] looks fine.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>  |  "Binaries may die
WWW: <http://www.cs.mu.oz.au/~fjh>  |   but source code lives forever"
PGP: finger fjh at 128.250.37.3        |     -- leaked Microsoft memo.