Mailing List Archive

5.002 beta things
Larry, Andy, all: thanks a whole bunch. Looks good on unicos&solaris (sun's
C compiler).

okay, enough of that. :)

MakeMaker: Is 5.08 going into 5.002? or do we patch up 5.06 with our pet
patches? I'm in favor of putting 5.08 into 5.002, mainly for the
alphabetized documentation that went into 5.07 :) If we stay with 5.06 then
I have a patch at the end of this note which should be added.

MakeMaker-5.08: If it goes into 5.002 it'll have to check for xsubpp
version 1.923 instead of 1.922. 1.923 is getting upset about the
-noprototype switch that 5.07 wants to feed to it :)

Smiley: :)

XSUB.h: Ug. Sun's C compiler complains about end-of-loop code not
reached. I'm using XSRETURN_UNDEF. I hate to ask (I know it's just another
suddenly-important discussion I missed), but why did the one-time loop have
to be added?

Configure: I think $path_sep should be colon for Unix-ish things.


Dean


-----

*** lib/ExtUtils/MakeMaker.pm Sat Nov 18 15:01:05 1995
--- ../patches/MakeMaker.pm Tue Nov 21 23:34:31 1995
***************
*** 960,969 ****
my($c); ($c = $name) =~ s/\.xs$/.c/;
$xs{$name} = $c;
$c{$c} = 1;
! } elsif ($name =~ /\.c$/){
$c{$name} = 1
unless $name =~ m/perlmain\.c/; # See MAP_TARGET
! } elsif ($name =~ /\.h$/){
$h{$name} = 1;
} elsif ($name =~ /\.(p[ml]|pod)$/){
$pm{$name} = $self->catfile('$(INST_LIBDIR)',"$name");
--- 960,969 ----
my($c); ($c = $name) =~ s/\.xs$/.c/;
$xs{$name} = $c;
$c{$c} = 1;
! } elsif ($name =~ /\.c$/i){
$c{$name} = 1
unless $name =~ m/perlmain\.c/; # See MAP_TARGET
! } elsif ($name =~ /\.h$/i){
$h{$name} = 1;
} elsif ($name =~ /\.(p[ml]|pod)$/){
$pm{$name} = $self->catfile('$(INST_LIBDIR)',"$name");
***************
*** 1039,1045 ****
$self->{PM} = \%pm unless $self->{PM};
$self->{C} = [sort keys %c] unless $self->{C};
my(@o_files) = @{$self->{C}};
! $self->{O_FILES} = [grep s/\.c$/$self->{OBJ_EXT}/, @o_files] ;
$self->{H} = [sort keys %h] unless $self->{H};
$self->{PL_FILES} = \%pl_files unless $self->{PL_FILES};

--- 1039,1045 ----
$self->{PM} = \%pm unless $self->{PM};
$self->{C} = [sort keys %c] unless $self->{C};
my(@o_files) = @{$self->{C}};
! $self->{O_FILES} = [grep s/\.c$/$self->{OBJ_EXT}/i, @o_files] ;
$self->{H} = [sort keys %h] unless $self->{H};
$self->{PL_FILES} = \%pl_files unless $self->{PL_FILES};

***************
*** 1740,1745 ****
--- 1740,1748 ----
push @m, '
.c$(OBJ_EXT):
$(CCCMD) $(CCCDLFLAGS) -I$(PERL_INC) $(DEFINE) $*.c
+
+ .C$(OBJ_EXT):
+ $(CCCMD) $(CCCDLFLAGS) -I$(PERL_INC) $(DEFINE) $*.C
';
join "", @m;
}
Re: 5.002 beta things [ In reply to ]
>>>>> "dean" == Dean Roehrich <roehrich@cray.com> writes:

dean> Larry, Andy, all: thanks a whole bunch. Looks good on unicos&solaris (sun's
dean> C compiler).

dean> okay, enough of that. :)

Not enough. Thanks, Thanks, Thanks, Larry, Andy, all...

100 tests o.k. on irix 5.3 and MakeMaker 5.08

dean> MakeMaker: Is 5.08 going into 5.002?

Yes, please! Most important: the AIX bug in MM 5.06 is really fixed
now (Thanks to Jens Peter Christensen for running the tests)

I've uploaded 5.08 to CPAN a few hours ago. (CPAN/authors/id/ANDK)

dean> MakeMaker-5.08: If it goes into 5.002 it'll have to check for xsubpp
dean> version 1.923 instead of 1.922. 1.923 is getting upset about the
dean> -noprototype switch that 5.07 wants to feed to it :)

dean> Smiley: :)

Done :-)

[...]

andreas
Re: 5.002 beta things [ In reply to ]
On Wed, 22 Nov 1995, Dean Roehrich wrote:

> XSUB.h: Ug. Sun's C compiler complains about end-of-loop code not
> reached. I'm using XSRETURN_UNDEF. I hate to ask (I know it's just another
> suddenly-important discussion I missed), but why did the one-time loop have
> to be added?

Tim did that. I'll have to go digging in my archives to remember why.
(Tim's currently out of town missing all the fun.)

Andy Dougherty doughera@lafcol.lafayette.edu
Re: 5.002 beta things [ In reply to ]
: On Wed, 22 Nov 1995, Dean Roehrich wrote:
:
: > XSUB.h: Ug. Sun's C compiler complains about end-of-loop code not
: > reached. I'm using XSRETURN_UNDEF. I hate to ask (I know it's just another
: > suddenly-important discussion I missed), but why did the one-time loop have
: > to be added?
:
: Tim did that. I'll have to go digging in my archives to remember why.

He was just copying me. It's a slightly well-known trick for writing
macros that parse like a statement in all circumstances, so you don't
have to remember whether to put a semicolon there or not. Most other
methods fail when you say things like this:

if (whatever)
XSRETURN_UNDEF;
else
XSRETURN_YES;

I would say Sun's C compiler needs to be fixed, or ignored.

Larry