[m-rev.] [reuse] for review: move type definitions to prog_data

[Zoltan Somogyi]

On 10-Jun-2004, Nancy Mazur <Nancy.Mazur at cs.kuleuven.ac.be> wrote:
> This is my plan for further adapting and improving the module dependencies of
> the alias and reuse modules: 

The five step plan is fine.

> Index: hlds_pred.m
> ===================================================================
> RCS file: /home/mercury1/repository/mercury/compiler/hlds_pred.m,v
> retrieving revision 1.81.2.22
> diff -u -r1.81.2.22 hlds_pred.m
> --- hlds_pred.m	2 Jun 2004 10:30:38 -0000	1.81.2.22
> +++ hlds_pred.m	10 Jun 2004 05:23:09 -0000
> @@ -23,7 +23,7 @@
>  :- import_module bool, list, set, map, std_util, term, varset.
>  :- import_module possible_alias.
>  :- import_module possible_alias__pa_alias_as.
> -:- import_module possible_alias__pa_datastruct.
> +% :- import_module possible_alias__pa_datastruct.
>  :- import_module structure_reuse.
>  :- import_module structure_reuse__sr_data.
>  
> @@ -1812,11 +1812,11 @@
>  :- mode proc_info_set_reuse_information(in, in, out) is det.
>  
>  :- pred proc_info_static_terms(proc_info, 
> -		maybe(set(pa_datastruct__datastruct))).
> +		maybe(set(prog_data__datastruct))).
>  :- mode proc_info_static_terms(in, out) is det.

If you didn't have the module qualifier on the type name, you wouldn't
have to update this. This we why usually omit module qualifiers on type names.

> Index: prog_data.m
> ===================================================================
> RCS file: /home/mercury1/repository/mercury/compiler/prog_data.m,v
> retrieving revision 1.58.2.16
> diff -u -r1.58.2.16 prog_data.m
> --- prog_data.m	2 Jun 2004 10:30:50 -0000	1.58.2.16
> +++ prog_data.m	10 Jun 2004 05:23:23 -0000
> @@ -22,6 +22,10 @@
>  % Any types which are needed in both the parse tree and in the HLDS
>  % should be defined here, rather than in hlds*.m.
>  
> +	% XXX temporary. Type "cons_id" needed for selector.
> +:- import_module hlds.
> +:- import_module hlds__hlds_data.
> +
>  :- import_module (parse_tree__inst), libs__options, libs__globals.
>  :- import_module recompilation.
>  :- import_module bool, list, assoc_list, map, set, varset, term, std_util.
> @@ -429,6 +433,28 @@
>  				% possible inputs.
>  	;	can_loop.	% This procedure might not terminate.
>  
> +%
> +% Stuff for the 'pa_alias_info' and 'sr_reuse_info' pragma's.
> +% 

It would be nice to find better names for those pragmas. What about simply
"possible_alias" and "structure_reuse"?

This section should refer readers to your thesis for documentation.

> +:- type selector == list(unit_sel).
> +:- type unit_sel 
> +	---> 	us(hlds_data__cons_id, int) ;  % normal selector
> +		ts(prog_data__type).		 % type selector

You don't want to include the module qualifier in the type names
unless there is a possibility of a type with that name being defined
in mor than one module, which isn't the case in either type here.

> +:- type datastruct 
> +	---> 	cel(
> +			dsvar::		prog_var, 
> +			dsselector:: 	selector
> +		).

Again, it would be nicer to have better name for this type. How about
"selected_cell"? The field names would then become sc_var and sc_selector.

> +:- type alias == pair(datastruct).
> +:- type aliases == list(alias).
> +:- type reuse_tuple
> +	---> 	unconditional
> +	; 	conditional(
> +		    reuse_nodes		:: set(datastruct),
> +		    live_headvars	:: set(datastruct), 
> +		    alias_headvars	:: aliases
> +		).

Some documentation of the meanings of the fields would be nice.

Apart from the above, the diff is fine.

Zoltan.
--------------------------------------------------------------------------
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
--------------------------------------------------------------------------