Still don't understand. Internal helper functions should be internally in which? From where you want call "foo(X)"? Where you would like to define "foo(X)"?<br><br><div class="gmail_quote">On Wed, Apr 9, 2008 at 11:22 AM, Matthew Dempsky <<a href="mailto:matthew@dempsky.org">matthew@dempsky.org</a>> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Because I want to be able to abstract away how to instantiate a<br>
module, and I want to be able to define internal helper functions to<br>
do it, and I want to be able to call them "foo(X)" instead of<br>
"new(foo, X)".<br>
<div><div></div><div class="Wj3C7c"><br>
On Wed, Apr 9, 2008 at 2:20 AM, Hynek Vychodil <<a href="mailto:vychodil.hynek@gmail.com">vychodil.hynek@gmail.com</a>> wrote:<br>
> UGH, static functions? Where is benefit to do it? What matter? I don't<br>
> understand.<br>
><br>
><br>
><br>
> On Wed, Apr 9, 2008 at 11:04 AM, Matthew Dempsky <<a href="mailto:matthew@dempsky.org">matthew@dempsky.org</a>><br>
> wrote:<br>
><br>
> ><br>
> ><br>
> ><br>
> > It would be nice to be able to define arbitrary static functions<br>
> > within an abstract module. Right now you can kludge around this by<br>
> > exploiting that local functions named 'new' are always static and<br>
> > write something like<br>
> ><br>
> > new(incr, X) -> X + 1.<br>
> ><br>
> > but it would be nicer to just write<br>
> ><br>
> > -static([incr/1]).<br>
> > incr(X) -> X + 1.<br>
> ><br>
> > Below is a very rudimentary patch that adds support for this syntax.<br>
> > It's basically only enough to compile and run very simple test cases<br>
> > and to demonstrate that implementing this does not seem too difficult.<br>
> > (One problem I'm already aware of is that a static function foo/1 and<br>
> > an instance function foo/0 silently collide without any warning... so<br>
> > don't do that. ;-))<br>
> ><br>
> > Another option would be to make the attribute statement just<br>
> > "-static([incr])." and then all incr functions are defined static, but<br>
> > that seems unnecessarily limiting.<br>
> ><br>
> > If the long-term goal is to have separate namespaces for static and<br>
> > instance functions (e.g., so (Mod:new(...)):module_info() works and<br>
> > returns something meaningful for the module instance), then maybe a<br>
> > syntax that actually decorates individual function definitions would<br>
> > be better?<br>
> ><br>
> > Are there any plans on adding static functions in a future release?<br>
> ><br>
> ><br>
> > --- stdlib/src/erl_parse.yrl.orig 2007-11-27 07:57:32.000000000<br>
> -0800<br>
> > +++ stdlib/src/erl_parse.yrl 2008-04-09 01:25:31.000000000 -0700<br>
> > @@ -627,6 +627,12 @@<br>
> > {attribute,La,export,farity_list(ExpList)};<br>
> > _Other -> error_bad_decl(La, export)<br>
> > end;<br>
> > +build_attribute({atom,La,static}, Val) -><br>
> > + case Val of<br>
> > + [ExpList] -><br>
> > + {attribute,La,static,farity_list(ExpList)};<br>
> > + _Other -> error_bad_decl(La, static)<br>
> > + end;<br>
> > build_attribute({atom,La,import}, Val) -><br>
> > case Val of<br>
> > [Name] -><br>
> > --- compiler/src/sys_pre_expand.erl.orig 2008-02-05<br>
> 05:37:09.000000000 -0800<br>
> > +++ compiler/src/sys_pre_expand.erl 2008-04-09 01:26:06.000000000<br>
> -0700<br>
> > @@ -38,6 +38,7 @@<br>
> > parameters=undefined, %Module parameters<br>
> > package="", %Module package<br>
> > exports=[], %Exports<br>
> > + statics=[], %Statics<br>
> > imports=[], %Imports<br>
> > mod_imports, %Module Imports<br>
> > compile=[], %Compile flags<br>
> > @@ -109,6 +110,7 @@<br>
> > end,<br>
> > {Fs1,Xs,Ds} = sys_expand_pmod:forms(Fs0, Ps,<br>
> > St0#expand.exports,<br>
> > + St0#expand.statics,<br>
> > St0#expand.defined),<br>
> > St1 = St0#expand{exports=Xs, defined=Ds},<br>
> > {Fs2,St2} = add_instance(Ps, Fs1, St1),<br>
> > @@ -220,6 +222,8 @@<br>
> > package = packages:strip_last(M)};<br>
> > attribute(export, Es, St) -><br>
> > St#expand{exports=union(from_list(Es), St#expand.exports)};<br>
> > +attribute(static, Ss, St) -><br>
> > + St#expand{statics=union(from_list(Ss), St#expand.statics)};<br>
> > attribute(import, Is, St) -><br>
> > import(Is, St);<br>
> > attribute(compile, C, St) when is_list(C) -><br>
> > --- compiler/src/sys_expand_pmod.erl.orig 2008-02-05<br>
> 05:38:12.000000000 -0800<br>
> > +++ compiler/src/sys_expand_pmod.erl 2008-04-09 01:32:36.000000000<br>
> -0700<br>
> > @@ -25,18 +25,19 @@<br>
> > %% and 'exports'. The automatic 'new/N' function is neither added to the<br>
> > %% definitions nor to the 'exports'/'defines' lists yet.<br>
> ><br>
> > --export([forms/4]).<br>
> > +-export([forms/5]).<br>
> ><br>
> > --record(pmod, {parameters, exports, defined, predef}).<br>
> > +-record(pmod, {parameters, exports, statics, defined, predef}).<br>
> ><br>
> > %% TODO: more abstract handling of predefined/static functions.<br>
> ><br>
> > -forms(Fs0, Ps, Es0, Ds0) -><br>
> > +forms(Fs0, Ps, Es0, Ss0, Ds0) -><br>
> > PreDef = [{module_info,0},{module_info,1}],<br>
> > - forms(Fs0, Ps, Es0, Ds0, PreDef).<br>
> > + forms(Fs0, Ps, Es0, Ss0, Ds0, PreDef).<br>
> ><br>
> > -forms(Fs0, Ps, Es0, Ds0, PreDef) -><br>
> > - St0 = #pmod{parameters=Ps,exports=Es0,defined=Ds0, predef=PreDef},<br>
> > +forms(Fs0, Ps, Es0, Ss0, Ds0, PreDef) -><br>
> > + St0 = #pmod{parameters=Ps,exports=Es0,statics=Ss0,defined=Ds0,<br>
> > + predef=PreDef},<br>
> > {Fs1, St1} = forms(Fs0, St0),<br>
> > Es1 = update_function_names(Es0, St1),<br>
> > Ds1 = update_function_names(Ds0, St1),<br>
> > @@ -50,15 +51,22 @@<br>
> > [update_function_name(E, St) || E <- Es].<br>
> ><br>
> > update_function_name(E={F,A}, St) when F =/= new -><br>
> > - case ordsets:is_element(E, St#pmod.predef) of<br>
> > + case ordsets:is_element(E, St#pmod.predef) orelse<br>
> > + ordsets:is_element(E, St#pmod.statics) of<br>
> > true -> E;<br>
> > false -> {F, A + 1}<br>
> > end;<br>
> > update_function_name(E, _St) -><br>
> > E.<br>
> ><br>
> > -update_forms([{function,L,N,A,Cs}|Fs],St) when N =/= new -><br>
> > - [{function,L,N,A+1,Cs}|update_forms(Fs,St)];<br>
> > +update_forms([F0={function,L,N,A,Cs}|Fs],St) when N =/= new -><br>
> > + F = case ordsets:is_element({N,A}, St#pmod.statics) of<br>
> > + true -><br>
> > + F0;<br>
> > + false -><br>
> > + {function,L,N,A+1,Cs}<br>
> > + end,<br>
> > + [F|update_forms(Fs,St)];<br>
> > update_forms([F|Fs],St) -><br>
> > [F|update_forms(Fs,St)];<br>
> > update_forms([],_St) -><br>
> > @@ -74,9 +82,14 @@<br>
> > {[], St0}.<br>
> ><br>
> > %% Only function definitions are of interest here. State is not updated.<br>
> > -form({function,Line,Name0,Arity0,Clauses0},St) when Name0 =/= new -><br>
> > - {Name,Arity,Clauses} = function(Name0, Arity0, Clauses0, St),<br>
> > - {{function,Line,Name,Arity,Clauses},St};<br>
> > +form(F = {function,Line,Name0,Arity0,Clauses0},St) when Name0 =/= new -><br>
> > + case ordsets:is_element({Name0, Arity0}, St#pmod.statics) of<br>
> > + true -><br>
> > + {F,St};<br>
> > + false -><br>
> > + {Name,Arity,Clauses} = function(Name0, Arity0, Clauses0, St),<br>
> > + {{function,Line,Name,Arity,Clauses},St}<br>
> > + end;<br>
> > %% Pass anything else through<br>
> > form(F,St) -> {F,St}.<br>
> ><br>
> > @@ -362,7 +375,13 @@<br>
> > expr({call,Lc,{atom,Lf,F},As0},St) -><br>
> > %% Local function call - needs THIS parameter.<br>
> > As1 = expr_list(As0,St),<br>
> > - {call,Lc,{atom,Lf,F},As1 ++ [{var,0,'THIS'}]};<br>
> > + As2 = case ordsets:is_element({F, length(As1)}, St#pmod.statics) of<br>
> > + true -><br>
> > + As1;<br>
> > + false -><br>
> > + As1 ++ [{var,0,'THIS'}]<br>
> > + end,<br>
> > + {call,Lc,{atom,Lf,F},As2};<br>
> > expr({call,Line,F0,As0},St) -><br>
> > %% Other function call<br>
> > F1 = expr(F0,St),<br>
> > _______________________________________________<br>
> > erlang-questions mailing list<br>
> > <a href="mailto:erlang-questions@erlang.org">erlang-questions@erlang.org</a><br>
> > <a href="http://www.erlang.org/mailman/listinfo/erlang-questions" target="_blank">http://www.erlang.org/mailman/listinfo/erlang-questions</a><br>
> ><br>
><br>
><br>
><br>
> --<br>
> --Hynek (Pichi) Vychodil<br>
</div></div></blockquote></div><br><br clear="all"><br>-- <br>--Hynek (Pichi) Vychodil