[Shake-dev] ShakeMap installation and user inquiries
Mitchell Gold
goldm at ldeo.columbia.edu
Mon Jul 16 17:17:07 UTC 2012
On 07/16/2012 12:41 PM, shake-dev-request at geohazards.usgs.gov wrote:
> Date: Mon, 16 Jul 2012 09:41:39 -0700
> From: Bruce Worden<cbworden at gmail.com>
> To: Branden Christensen<branden.christensen at osop.com.pa>
> Cc: shake-dev at geohazards.usgs.gov
> Subject: Re: [Shake-dev] ShakeMap installation and user inquiries
> Message-ID:<65D369D7-11BB-46F8-A97D-36BA59AF501C at gmail.com>
>
> Congratulations! I'm glad you've got it working.
>
> I'm a bit mystified as to why these changes were necessary. Most systems work fine with the supplied version. I suspect there is some default behavior of some system program (like 'mv' or 'cp') or library function that is different on Ubuntu than other systems (or you have a new version of a Perl module that changed something similar). If you can point to the precise line where the error occurred, I'll try to track it down.
>
> Somehow you've managed to avoid the usual problems with Ubuntu, which occur when the C programs in the "contour" directory try to link with GMT and NetCDF. Those problems were, in the past, seemingly insurmountable. At least none of the several people who tried ever solved it.
>
> Good luck, and let us know about your progress.
>
> Bruce
Bruce,
I have been struggling with similar issues for a while on our Centos 5.5
Intel 64-bit system. I noticed the errors were due to sm_nearneighbor
not compiling in the contour directory and more specifically that things
there are incompatible with libgmt.a which ends up mostly being compiled
-m32 in GMT (src/config.mk: CFLAGS, CPPFLAGS). I recompiled GMT without
the -m32 and took the resulting libgmt.a which now had all 64-bit object
modules for the contour compile, and it appears as if everything needed
is working so far.
--Mitch Gold
Lamont-Doherty Cooperative Seismographic Network
goldm at ldeo.columbia.edu
845-365-8583
More information about the Shake-dev
mailing list