[m-rev.] for review: postpone the processing of reuse/alias-pragma's

[Nancy Mazur]

Hi Peter, 

I'll do the suggested changes, but I have some remarks... 

> > +:- pred mode_list_matches(list(mode), list(mode), module_info).
> > +:- mode mode_list_matches(in, in, in) is semidet.
> > +
> > +mode_list_matches([], [], _).
> > +mode_list_matches([Mode1 | Modes1], [Mode2 | Modes2], ModuleInfo) :-
> > +	% Use mode_get_insts_semidet instead of mode_get_insts to avoid
> > +	% aborting if there are undefined modes.
> > +	mode_get_insts_semidet(ModuleInfo, Mode1, Inst1, Inst2),
> > +	mode_get_insts_semidet(ModuleInfo, Mode2, Inst1, Inst2),
> > +	mode_list_matches(Modes1, Modes2, ModuleInfo).
> > +
> 
> It would be better to move this mode_list_matches predicate into
> mode_util.m and delete it from make_hlds.m.

perfect ! ;-) 

> > +process_unproc_alias_pragma(UnprocAliasPragma, Module0, Module) --> 
> > +	{ UnprocAliasPragma = unproc_alias_pragma(PredOrFunc, SymName, 
> > +		Modes, HeadVars, Types, Alias0) },
> > +
> > +	globals__io_lookup_bool_option(very_verbose, VeryVerbose),
> > +
> > +	{ module_info_get_predicate_table(Module0, Preds) }, 
> > +	{ list__length(Modes, Arity) },
> > +	(
> > +		{ predicate_table_search_pf_sym_arity_declmodes(Module0, 
> > +			Preds, PredOrFunc, SymName, Arity, Modes, 
> > +			PredId, ProcId) }
> > +	->
> > +		{ module_info_preds(Module0, PredTable0) },
> > +		{ map__lookup(PredTable0, PredId, PredInfo0) },
> > +		{ pred_info_procedures(PredInfo0, ProcTable0) },
> > +		{ map__lookup(ProcTable0, ProcId, ProcInfo0) },
> > +		write_proc_progress_message("(Alias) Looking into ", 
> > +			PredId, ProcId, Module0),
> > +
> > +			% rename the headvars: 
> > +		maybe_write_string(VeryVerbose, "Renaming HeadVars..."),
> > +		{ proc_info_headvars(ProcInfo0, ProcHeadVars) }, 
> > +		{ list__map(term__coerce_var, HeadVars, CHVars) },
> > +		{ map__from_corresponding_lists(CHVars, ProcHeadVars,
> > +			MapHeadVars) }, 
> > +		{ pa_alias_as__rename(MapHeadVars, Alias0, Alias1) },
> > +		maybe_write_string(VeryVerbose, "done.\n"),
> > +	
> > +		% rename the types: 
> > +		maybe_write_string(VeryVerbose, "Renaming Types..."),
> > +		{ pred_info_arg_types(PredInfo0, ArgTypes) },
> > +		{ pa_alias_as__rename_types(Types, ArgTypes, 
> > +			Alias1, Alias) },
> > +		maybe_write_string(VeryVerbose, "done.\n"),
> > +
> > +		% set the proc_info right
> > +		{ proc_info_set_possible_aliases(ProcInfo0, 
> > +			Alias, ProcInfo) },
> > +		{ map__det_update(ProcTable0, ProcId, ProcInfo,
> > +				ProcTable) },
> > +		{ pred_info_set_procedures(PredInfo0, ProcTable,
> > +				PredInfo) },
> > +		{ map__det_update(PredTable0, PredId, PredInfo,
> > +				PredTable) },
> > +		{ module_info_set_preds(Module0, PredTable, Module) }
> > +	;
> > +		% io__write_string("Warning: no entry found for "),
> > +		% hlds_out__write_simple_call_id(PredOrFunc, SymName/Arity),
> > +		% io__write_string(" with modes: "), 
> > +		% { varset__init(EmptyVarset) },
> > +		% io__write_list(Modes, ", ", write_mode(EmptyVarset)),
> > +		% io__write_string(" (alias_info).\n"),
> 
> You definitely need to report an error here.
> 
> Use
> 
> io__set_exit_status(1),
> { module_info_incr_errors(Module0, Module) },
> 
> This should ensure that the compiler finishes with an abort.

I'm not so sure here. When uncommented, I obtain a huge list of such
warnings, but they don't seem to be worrying. Is it possible that some
of the predicates/procedures are removed from the hlds if known that
they are not needed? Hmm, stupid remark from me ;-) 
No, I think I know where the problem is. Given the fact that we have
our own system of transitively importing all trans_opts of
all the modules in which some of the needed types are defined, this
means that a lot of "unnecessary" pragma's are read. Pragma's which will
never be needed, because the actual procedures are never needed. 

Does this make sense? I think so.. 

Same remark for the reuse_info.. 


I'll process the rest of your comments and send a diff later on... 
Thanks! 
Nancy
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------