[m-dev.] Using VC++ 6 with hlc.gc

[Peter Ross]

The next change that I will be working on, is to use the
Microsoft extensions .obj, .lib and so on.

This will be quite a big change and mainly trivial stuff,
so I intend to check this change in as soon as I am sure
that it bootchecks under linux.

I will post a diff soon.

Pete
----- Original Message -----
From: "Fergus Henderson" <fjh at cs.mu.OZ.AU>
To: <mercury-developers at cs.mu.OZ.AU>
Sent: Tuesday, June 06, 2000 3:48 PM
Subject: Re: [m-dev.] Using VC++ 6 with hlc.gc


> On 06-Jun-2000, Peter Ross <peter.ross at miscrit.be> wrote:
> > Here it is!
>
> Great -- thanks.
>
> I have some preliminary review comments.
> These are mostly just minor style issues.
>
> > +++ boehm_gc/if_not_there.c 2000/06/06 10:10:24
> > @@ -16,8 +16,7 @@
> >          return(0);
> >      }
> >      printf("^^^^Starting command^^^^\n");
> > -    execvp(argv[2], argv+2);
> > -    exit(1);
> > +    return execvp(argv[2], argv+2);
> >
>
> I don't quite understand the rationale for that change.
> I guess it is a work-around for some brokenness in
> the MSVC implementation of execvp().
> If so, that change should be inside `#ifdef _MSC_VER'.
>
This change wasn't required after all because I should be using NT_MAKEFILE
to compile the library, not the unix makefile which uses if_not_there.c

> > +++ compiler/mercury_compile.m 2000/06/06 10:11:41
> > @@ -2696,8 +2696,9 @@
> >      ;
> >  TraceOpt = ""
> >      },
> > -     join_module_list(Modules, ".c", ["> ", InitCFileName],
MkInitCmd0),
> > -     { string__append_list(["c2init ", TraceOpt | MkInitCmd0],
> > +     join_module_list(Modules, ".c", ["> ", InitCFileName, "'"],
> > + MkInitCmd0),
> > +     { string__append_list(["bash -c 'c2init ", TraceOpt | MkInitCmd0],
> >      MkInitCmd) },
> >      invoke_system_command(MkInitCmd, MkInitOK),
>
> It would probably be better to change `invoke_system_command', or to
> add a new routine `invoke_shell_command', rather than changing the
> code here.  It would also be better to avoid hard-coding this; you
> could make it conditional on an option, which would be passed by
> scripts/mmc, with the default value passed by scripts/mmc for that
> option being determined by some autoconf variable set at configuration
> time.
>

I have defined a new predicate invoke_shell_command which uses a
function use_win32 which is true if we are compiling in a win32 environment
to decide whether to wrap any system commands with calls to
bash -c '<command>'.

> > Index: compiler/mlds_to_c.m
> > +:- pred mlds_needs_initialization(mlds__initializer::in, bool::out) is
det.
> > +
> > +mlds_needs_initialization(no_initializer, no).
> > +mlds_needs_initialization(init_obj(_), yes).
> > +mlds_needs_initialization(init_struct([]), no).
> > +mlds_needs_initialization(init_struct([_|_]), yes).
> > +mlds_needs_initialization(init_array(_), yes).
>
> I suggest making that a function rather than a predicate.
>
Done.

> > +++ scripts/Mmake.vars.in 2000/06/06 10:14:21
> > @@ -179,9 +179,11 @@
> >  MSPFLAGS =
> >  EXTRA_MSPFLAGS =
> >
> > -AR = ar
> > -ALL_ARFLAGS = $(ARFLAGS) $(EXTRA_ARFLAGS) $(TARGET_ARFLAGS)
> > -ARFLAGS = cr
> > + # Note ARFLAGS at the end as MS VC++6 expects a command line of the
> > + # form: lib /OUT:libxxx.a *.o
> > +AR = @AR@
> > +ALL_ARFLAGS = $(EXTRA_ARFLAGS) $(TARGET_ARFLAGS) $(ARFLAGS)
> > +ARFLAGS = @ARFLAGS@
>
> I think it would be clearer to introduce a different variable
> called say AR_LIBFILE_OPT, similar to the OBJFILE_OPT variable
> that you introduced to use instead of `-o' for the C compiler,
> and to use that rather than changing the order of $(ARFLAGS)
> in $(ALL_ARFLAGS) and requiring that there be no space
> between $(ALL_ARFLAGS) and the library name.
>
Done.

> > +++ trace/Mmakefile 2000/06/06 10:14:30
> > @@ -124,7 +124,7 @@
> >
> >  lib$(TRACE_LIB_NAME)$(DLL_DEF_LIB).a: $(OBJS)
> >  rm -f lib$(TRACE_LIB_NAME)$(DLL_DEF_LIB).a
> > - ar cr lib$(TRACE_LIB_NAME)$(DLL_DEF_LIB).a $(OBJS)
> > + $(AR) $(ARFLAGS)lib$(TRACE_LIB_NAME)$(DLL_DEF_LIB).a $(OBJS)
>
> Here (and elsewhere) that should probably be "$(ALL_ARFLAGS)"
> or perhaps "$(ALL_ARFLAGS) $(AR_LIBFILE_OPT)" rather
> than "$(ARFLAGS)".
>
> > +++ util/mkinit.c 2000/06/06 10:14:39
> > @@ -23,8 +23,12 @@
> ...
> > +#ifdef HAVE_UNISTD_H
> > +#include <unistd.h>
> > +#endif
>
> It's nicer to use indentation to indicate the nesting:
>
> s/#include/  #include/
>            ^^
>
> > +++ runtime/mercury_memory.c 2000/06/06 10:15:35
> > +++ runtime/mercury_memory_handlers.c 2000/06/06 10:15:37
> > +++ runtime/mercury_memory_zones.c 2000/06/06 10:15:40
> > +++ runtime/mercury_prof.c 2000/06/06 10:15:42
> > +++ runtime/mercury_reg_workarounds.c 2000/06/06 10:15:42
> > +++ runtime/mercury_signal.c 2000/06/06 10:15:42
> > +++ runtime/mercury_timing.c 2000/06/06 10:15:43
> > +++ runtime/mercury_timing.h 2000/06/06 10:15:43
> > +++ runtime/mercury_trace_base.c 2000/06/06 10:15:47
>
> Likewise.
>
Done.

> > +++ runtime/mercury_reg_workarounds.h 2000/06/06 10:15:42
> > @@ -18,6 +18,8 @@
> >    #include <sys/time.h> /* for FD_ZERO() */
> >  #endif
> >
> > +#include <stdlib.h> /* needed for size_t with VC++ 6 */
>
> I think it would be sufficient to say "for size_t".
> The fact that it works without that #include on other systems
> is just coincidence; it might well break on future releases
> of Linux, for example.
>
Done.

> --
> Fergus Henderson <fjh at cs.mu.oz.au>  |  "I have always known that the
pursuit
> WWW: <http://www.cs.mu.oz.au/~fjh>  |  of excellence is a lethal habit"
> PGP: finger fjh at 128.250.37.3        |     -- the last words of T. S. Garp.
> --------------------------------------------------------------------------
> mercury-developers mailing list
> Post messages to:       mercury-developers at cs.mu.oz.au
> Administrative Queries: owner-mercury-developers at cs.mu.oz.au
> Subscriptions:          mercury-developers-request at cs.mu.oz.au
> --------------------------------------------------------------------------

--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------