[m-rev.] for review: mmc --make, submodules and java

[Julien Fischer]

On Tue, 5 May 2009, Peter Wang wrote:

> 2009/5/5 Julien Fischer <juliensf at csse.unimelb.edu.au>:
>>> -:- import_module libs.globals.
>>> -
>>> :- import_module libs.compiler_util.
>>> +:- import_module libs.globals.
>>> :- import_module libs.handle_options.
>>> :- import_module libs.options.
>>> +:- import_module mdbcomp.prim_data.
>>> +:- import_module ml_backend.                % XXX unwanted dependency
>>> +:- import_module ml_backend.mlds_to_java.   % XXX unwanted dependency
>>> :- import_module parse_tree.prog_util.
>>> :- import_module parse_tree.source_file_map.
>>
>> I strongly suggest not (re)introducing this dependency.  The alternative,
>> duplicating a little code, seems preferable in this case.  (You presumably
>> can't factor out the name mangling code into the libs package because
>> it depends on the mlds_sym_name type.)
>
> It only used mlds_qual_kind, which only says if the initial letter should be
> case flipped or not.  There should be no problem factoring it out.  It calls
> prog_foreign.name_mangle though, so either it goes into parse_tree, or we
> move name_mangle into libs as well.  I think libs is preferred.

parse_tree is fine; it's introducing import dependencies back the other way
that concerns me, i.e. having the frontend packages import the code
generators.

Julien.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------