This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [m3user] Issues on installing CMAQ5.0 :unrecognized command line option "-finit-character=32"
- From: Tom Roche <Tom_Roche at pobox dot com>,
- To: m3user at listserv dot unc dot edu,
- Cc: fortran at gcc dot gnu dot org,
- Date: Sun, 24 Jun 2012 18:22:08 -0400
- Subject: Re: [m3user] Issues on installing CMAQ5.0 :unrecognized command line option "-finit-character=32"
- References: <4FE72BF3.1080309@net-b.de> <LYRIS-24886807-31705255-2012.06.23-16.32.38--tom_roche#pobox.com@listserv.unc.edu> <LYRIS-24886807-31705056-2012.06.23-14.22.58--tom_roche#pobox.com@listserv.unc.edu> <LYRIS-42101171-31604775-2012.06.04-01.52.19--yaoxianh#mtu.edu@listserv.unc.edu> <87ehp5g3vn.fsf@pobox.com>
- Reply-to: m3user at listserv dot unc dot edu, Tom Roche <Tom_Roche at pobox dot com>
Yaoxian Huang Sat, 23 Jun 2012 14:22:48 -0400
>>> I also met the exact same problem
...
>>>> After netCDF,I/O API 3.1 and MPICH has been installed, I tried to
>>>> compile the main CMAQ5.0 by excuting './bldit.bldmake' in
>>>> $M3HOME/scripts/build directory, but something was wrong:
>>>> gfortran -c -ffixed-form -ffixed-line-length-132 -O3 -funroll-loops -finit-character=32 cfg_module.F
>>>> cc1: error:unrecognized command line option "-finit-character=32"
Tobias Burnus Sun, 24 Jun 2012 17:02:11 +0200
> The flag -finit-character=32 is supported since GCC 4.3. The flag
> initializes local character variables with blank, while according to
> the standard those variables have an undefined value; given that
> this is a rather special flag, my suspicion is that it also works
> without that flag. However, it probably makes more sense to use a
> newer GCC version.
>>>> Platform : RH5.5 AS 32bit
Tom Roche Sat, 23 Jun 2012 16:38:52 -0400
>> I suspect you have a downlevel gfortran[: see]
the new'n'improved
http://www.cmaq-model.org/cmaqwiki/index.php?title=CMAQ_version_5.0_%28February_2010_release%29_OGD#GCC
>> You can dis/prove my hypothesis by running
>> $ gfortran --version
>> If correct, how to fix? One way is to upgrade your RHEL (the
>> current version==6), but that may be outside your control. If so, I
>> suspect the fine folks on the gfortran list can suggest the best
>> version of GCC to install on RHEL 5.5.
> Red Hat Enterprise Server 5.x uses GCC 4.1.x as system compiler.
> However, Red Hat provides as technical preview a never GCC version.
> I think "yum install gcc44-gfortran" (as root) should install the
> technical preview along the default GCC. I assume - but I don't know
> - that that compiler is then available as "gfortran44",
> "gfortran-4.4" or something similar.
> (I think RH 5.5 has 4.4 as technical preview, but it could also have
> gcc43-gfortran, I don't recall which RHES 5.x had which version as
> technical preview. An overview about which distribution has which
> GCC version is at
> http://gcc.gnu.org/wiki/GFortranDistros
rearranged
> Alternatively, you could install an *unofficial* GCC build
http://gcc.gnu.org/wiki/GFortranBinaries#GNU.2BAC8-Linux
> or compile GCC yourself
http://gcc.gnu.org/wiki/GFortranBinaries#FromSource
Thanks, I have added links to those from the CMAQ wikipage (above).
> For completeness: GCC 4.3 was released in March 2008, more than 4
> years ago. The GCC projects supports compiler [versions] for about
> two years; currently supported are GCC 4.5 (4.5.3 is the latest
> version), 4.6 (4.6.3) and 4.7 (4.7.1) - and the current developer
> version 4.8, which is usually quite stable/usable and which will be
> released around March/April 2013.