Hi all,

I'm trying to build PostgreSQL 8.1.2 on Solaris 10 U1 (x86). Here's
the configure command I used:

#!/bin/ksh

export CC=cc
export CCC=CC
export CFLAGS=-O

cd postgres
LDFLAGS=-R/usr/sfw/lib ./configure \
--prefix=/opt/local \
--with-openssl \
--with-includes=/usr/sfw/include \
--with-libraries=/usr/sfw/lib \
--without-readline
cd ..

The setting of LDFLAGS is required because the default doesn't include
-R flags, even though the docs say that "--enable-rpath" is the default
(and therefore implies that it should do this).

But to the point: after a successful "configure" run, running gmake
(from /usr/sfw/bin) results in a core dump. Here's the last few lines
from truss:

stat64("GNUmakefile.in", 0x08046C10) = 0
brk(0x08093A58) = 0
brk(0x08095A58) = 0
sigaction(SIGABRT, 0x00000000, 0x08046BD0) = 0
sigaction(SIGABRT, 0x08046B10, 0x08046B90) = 0
schedctl() = 0xFEDD1000
lwp_sigmask(SIG_SETMASK, 0x00000000, 0x00000000) = 0xFFBFFEFF [0x0000FFFF]
lwp_kill(1, SIGABRT) = 0
Received signal #6, SIGABRT [default]
siginfo: SIGABRT pid=13033 uid=1001 code=-1

Does anyone know what's going on here, and more to the point, has
anyone successfuly compiled Postgres 8.1.2 on Solaris 10 U1? I'm
using Sun's Studio 11 compiler (of course), but I don't think that
matters at the moment because I don't think gmake is getting as far
as running a compiler.

TIA,

--
Rich Teer, SCNA, SCSA, OpenSolaris CAB member

President,
Rite Online Inc.

Voice: +1 (250) 979-1638
URL: http://www.rite-group.com/rich

On 1/2/06 12:11, in article Pine.SOL.4.64.0601311558490.4292@marrakesh,
"Rich Teer" <rich.teer@rite-group.com> wrote:

> Hi all,
>
> I'm trying to build PostgreSQL 8.1.2 on Solaris 10 U1 (x86). Here's
> the configure command I used:
>
> #!/bin/ksh
>
> export CC=cc
> export CCC=CC
> export CFLAGS=-O
>
> cd postgres
> LDFLAGS=-R/usr/sfw/lib ./configure \
> --prefix=/opt/local \
> --with-openssl \
> --with-includes=/usr/sfw/include \
> --with-libraries=/usr/sfw/lib \
> --without-readline
> cd ..
>
> The setting of LDFLAGS is required because the default doesn't include
> -R flags, even though the docs say that "--enable-rpath" is the default
> (and therefore implies that it should do this).
>
> But to the point: after a successful "configure" run, running gmake
> (from /usr/sfw/bin) results in a core dump. Here's the last few lines
> from truss:
>
> stat64("GNUmakefile.in", 0x08046C10) = 0
> brk(0x08093A58) = 0
> brk(0x08095A58) = 0
> sigaction(SIGABRT, 0x00000000, 0x08046BD0) = 0
> sigaction(SIGABRT, 0x08046B10, 0x08046B90) = 0
> schedctl() = 0xFEDD1000
> lwp_sigmask(SIG_SETMASK, 0x00000000, 0x00000000) = 0xFFBFFEFF [0x0000FFFF]
> lwp_kill(1, SIGABRT) = 0
> Received signal #6, SIGABRT [default]
> siginfo: SIGABRT pid=13033 uid=1001 code=-1
>
> Does anyone know what's going on here, and more to the point, has
> anyone successfuly compiled Postgres 8.1.2 on Solaris 10 U1? I'm
> using Sun's Studio 11 compiler (of course), but I don't think that
> matters at the moment because I don't think gmake is getting as far
> as running a compiler.

There are packages for Postgres 8.1.2 in Blastwave's "testing" directory, so
obviously they're buildable *somehow*. You could either ask the Blastwave
maintainer for the magic, or just grab those packages.

<http://www.blastwave.org/testing/>

Cheers,

Chris