Re: [PATCH 1/5] pkl: Postpone decision of completeness of a struct literals

[Mohammad-Reza Nabipoor]

On Sun, Jan 23, 2022 at 07:03:55PM +0100, Jose E. Marchesi wrote:
> 
> > Hi, Jose
> >> >    PKL_AST_TYPE (node) = ASTREF (type);
> >> > -  PKL_AST_TYPE_COMPLETE (type) = type_complete;
> >> > +  PKL_PASS_RESTART = 1;
> >> 
> >> Why this restart?
> >> 
> >
> > We're creating new nodes using `pkl_ast_make_struct_type_field`.
> > The code above is like this:
> >
> > ```c
> >       pkl_ast_node struct_type_field
> >         = pkl_ast_make_struct_type_field (PKL_PASS_AST,
> >                                           PKL_AST_STRUCT_FIELD_NAME (t),
> >                                           PKL_AST_TYPE (t),
> >                                           NULL /* constraint */,
> >                                           NULL /* initializer */,
> >                                           NULL /* label */,
> >                                           PKL_AST_ENDIAN_DFL /* endian */,
> >                                           NULL /* optcond */);
> >
> >       struct_field_types = pkl_ast_chainon (struct_field_types,
> >                                             struct_type_field);
> > ```
> >
> > I think thse new fields should be processed.
> 
> You are right :)
> 

Pushed.
Thanks.