On Dec 20, 2007, at 04:17, Takeshi Enomoto wrote:
Would it be OK with the maintainer if we just commit that change now?
I compile libnc-dap with g95. I don't use gfortran. I don't have any opinion which version of gcc4.x for libnc-dap.
Having said that, gcc43 should build OK on Tiger since it is MacPorts policy to support the current and previous OS versions as far as I know.
No, we surely do not want to use gcc43, because that is not a final version. We want to use gcc42, unless we can prove that the port does not work with gcc42 but would work with gcc43. But I doubt that's the case.
BTW, a fix for g95 has not been commited. Could you take a look?
I'll have a look.
gcc42 installs for me, but gcc40 does not. Is there a way to force the build of stellarium and others, to use gcc42 instead of the default gcc40 ? On 12/20/07 7:03 AM, Ryan Schmidt wrote:
On Dec 20, 2007, at 04:17, Takeshi Enomoto wrote:
Would it be OK with the maintainer if we just commit that change now?
I compile libnc-dap with g95. I don't use gfortran. I don't have any opinion which version of gcc4.x for libnc-dap.
Having said that, gcc43 should build OK on Tiger since it is MacPorts policy to support the current and previous OS versions as far as I know.
No, we surely do not want to use gcc43, because that is not a final version. We want to use gcc42, unless we can prove that the port does not work with gcc42 but would work with gcc43. But I doubt that's the case.
BTW, a fix for g95 has not been commited. Could you take a look?
I'll have a look.
-- --- Barry McInnes 325 Broadway Boulder CO 80304 (303)4976231 barry.j.mcinnes@noaa.gov ---
This was in 10.4.x but not it seems in 10.5.x, unless its hidden somewhere ? 10.4.11 [mac05:~] bmcinnes% gcc_select Current default compiler: gcc version 4.0.1 (Apple Computer, Inc. build 5367) [mac05:~] bmcinnes% which gcc_select /usr/sbin/gcc_select 10.5.1 sh-3.2# man gcc_select No manual entry for gcc_select sh-3.2# which gcc_select sh-3.2# /usr/sbin/gcc_select sh: /usr/sbin/gcc_select: No such file or directory sh-3.2# On 12/20/07 1:07 PM, Ludovic Hirlimann wrote:
On 20 déc. 07, at 20:20, Barry McInnes wrote:
gcc42 installs for me, but gcc40 does not. Is there a way to force the build of stellarium and others, to use gcc42 instead of the default gcc40 ?
gcc_select ?
Ludo_______________________________________________ macports-users mailing list macports-users@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo/macports-users
-- --- Barry McInnes 325 Broadway Boulder CO 80304 (303)4976231 barry.j.mcinnes@noaa.gov ---
On Dec 20, 2007, at 9:07 PM, Ludovic Hirlimann wrote:
On 20 déc. 07, at 20:20, Barry McInnes wrote:
gcc42 installs for me, but gcc40 does not. Is there a way to force the build of stellarium and others, to use gcc42 instead of the default gcc40 ?
gcc_select ?
Nah, this won't have an effect on most ports (which doesn't mean it's a good idea to use it). In a perfect world (tm), the port will react to: `port install foo configure.compiler=FOO' where FOO might be something like "gcc-4.0" (default on 10.4 and 10.5), "gcc-3.3", "macports-gcc-4.2", etc. See [1] for all available options. (and of course even if this succeeds, it's still a bug) Regards, -Markus PS: Does stellarium really require the gcc 4.0 from MacPorts? [1] http://guide.macports.org/#reference.phases.configure -- Dipl. Inf. (FH) Markus W. Weissmann http://www.macports.org/ http://www.mweissmann.de/
On Dec 20, 2007, at 17:09, Markus Weissmann wrote:
On Dec 20, 2007, at 9:07 PM, Ludovic Hirlimann wrote:
On 20 déc. 07, at 20:20, Barry McInnes wrote:
gcc42 installs for me, but gcc40 does not. Is there a way to force the build of stellarium and others, to use gcc42 instead of the default gcc40 ?
gcc_select ?
Nah, this won't have an effect on most ports (which doesn't mean it's a good idea to use it). In a perfect world (tm), the port will react to: `port install foo configure.compiler=FOO' where FOO might be something like "gcc-4.0" (default on 10.4 and 10.5), "gcc-3.3", "macports-gcc-4.2", etc. See [1] for all available options. (and of course even if this succeeds, it's still a bug)
In a perfect world, ports would be updated to no longer depend on the gcc40 port, and instead depend on the gcc42 port, if appropriate.
PS: Does stellarium really require the gcc 4.0 from MacPorts?
The stellarium port does not depend on any gcc port. On first glance, there does not appear to be any gcc port anywhere in the dependency tree of stellarium (unless it occurs via a variant).
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I installed the dependencies one at a time, and installed gcc42 as well just to see if it would build which it does. Currently the stellarium error is sh-3.2# port install stellarium - ---> Staging boost into destroot Error: Target org.macports.destroot returned: no files matched glob pattern "*-1_34_1.dylib" Error: The following dependencies failed to build: boost Error: Status 1 encountered during processing. sh-3.2# port install boost - ---> Staging boost into destroot Error: Target org.macports.destroot returned: no files matched glob pattern "*-1_34_1.dylib" Error: Status 1 encountered during processing. sh-3.2# grads fails on gcc40 /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_gcc40/work/gcc-4.0.4/gcc/config/rs6000/host-darwin.c:130: warning: passing argument 1 of 'sigaltstack' from incompatible pointer type make[2]: *** [host-darwin.o] Error 1 make[2]: *** Waiting for unfinished jobs.... make[1]: *** [stage1_build] Error 2 make: *** [profiledbootstrap] Error 2 Error: The following dependencies failed to build: libnc-dap gcc40 Error: Status 1 encountered during processing. [mac27:grib2/netcdf-3.6.2/libsrc] root# gcc40 fails /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_gcc40/work/gcc-4.0.4/gcc/config/rs6000/host-darwin.c: In function 'darwin_rs6000_extra_signals': /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_gcc40/work/gcc-4.0.4/gcc/config/rs6000/host-darwin.c:130: warning: passing argument 1 of 'sigaltstack' from incompatible pointer type make[2]: *** [host-darwin.o] Error 1 make[2]: *** Waiting for unfinished jobs.... echo timestamp > s-gtype make[1]: *** [stage1_build] Error 2 make: *** [profiledbootstrap] Error 2 Error: Status 1 encountered during processing. On 12/21/07 1:29 AM, Ryan Schmidt wrote:
On Dec 20, 2007, at 17:09, Markus Weissmann wrote:
On Dec 20, 2007, at 9:07 PM, Ludovic Hirlimann wrote:
On 20 déc. 07, at 20:20, Barry McInnes wrote:
gcc42 installs for me, but gcc40 does not. Is there a way to force the build of stellarium and others, to use gcc42 instead of the default gcc40 ?
gcc_select ?
Nah, this won't have an effect on most ports (which doesn't mean it's a good idea to use it). In a perfect world (tm), the port will react to: `port install foo configure.compiler=FOO' where FOO might be something like "gcc-4.0" (default on 10.4 and 10.5), "gcc-3.3", "macports-gcc-4.2", etc. See [1] for all available options. (and of course even if this succeeds, it's still a bug)
In a perfect world, ports would be updated to no longer depend on the gcc40 port, and instead depend on the gcc42 port, if appropriate.
PS: Does stellarium really require the gcc 4.0 from MacPorts?
The stellarium port does not depend on any gcc port. On first glance, there does not appear to be any gcc port anywhere in the dependency tree of stellarium (unless it occurs via a variant).
- -- - --- Barry McInnes 325 Broadway Boulder CO 80304 (303)4976231 barry.j.mcinnes@noaa.gov - --- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (Darwin) iD8DBQFHbAsDKoTeRnsNi5kRAqGMAKDENGqrzZT2NmleoaCaDejDwHJOqACg0MEB 39D9EokzYiapuYubGNUo0pU= =lY/K -----END PGP SIGNATURE-----
On Dec 21, 2007, at 12:50, Barry McInnes wrote:
I installed the dependencies one at a time, and installed gcc42 as well just to see if it would build which it does.
Currently the stellarium error is
sh-3.2# port install stellarium - ---> Staging boost into destroot Error: Target org.macports.destroot returned: no files matched glob pattern "*-1_34_1.dylib" Error: The following dependencies failed to build: boost Error: Status 1 encountered during processing. sh-3.2# port install boost - ---> Staging boost into destroot Error: Target org.macports.destroot returned: no files matched glob pattern "*-1_34_1.dylib" Error: Status 1 encountered during processing. sh-3.2#
That is not a stellarium error, that is a boost error. Please file a bug against the boost port and make sure to Cc the maintainer of the boost port.
grads fails on gcc40
/opt/local/var/macports/build/ _opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_ gcc40/work/gcc-4.0.4/gcc/config/rs6000/host-darwin.c:130: warning: passing argument 1 of 'sigaltstack' from incompatible pointer type make[2]: *** [host-darwin.o] Error 1 make[2]: *** Waiting for unfinished jobs.... make[1]: *** [stage1_build] Error 2 make: *** [profiledbootstrap] Error 2
Error: The following dependencies failed to build: libnc-dap gcc40 Error: Status 1 encountered during processing. [mac27:grib2/netcdf-3.6.2/libsrc] root#
gcc40 fails
/opt/local/var/macports/build/ _opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_ gcc40/work/gcc-4.0.4/gcc/config/rs6000/host-darwin.c: In function 'darwin_rs6000_extra_signals': /opt/local/var/macports/build/ _opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_ gcc40/work/gcc-4.0.4/gcc/config/rs6000/host-darwin.c:130: warning: passing argument 1 of 'sigaltstack' from incompatible pointer type make[2]: *** [host-darwin.o] Error 1 make[2]: *** Waiting for unfinished jobs.... echo timestamp > s-gtype make[1]: *** [stage1_build] Error 2 make: *** [profiledbootstrap] Error 2
Error: Status 1 encountered during processing.
I'm not surprised gcc40 is failing to install. All ports that depend on gcc40 should be updated to depend on gcc42. I've been saying this for months. Please file bugs against any remaining ports which still depend on gcc40, and be sure to Cc those ports' maintainers, so that they can be updated. For unmaintained ports, please alert this list.
participants (4)
-
Barry McInnes
-
Ludovic Hirlimann
-
Markus Weissmann
-
Ryan Schmidt