This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
Hugh, A netCDF library can be built that supports large-files under Solaris 2.6. The trick is to ensure that the environment variables CFLAGS, LDFLAGS, and LIBS contain the necessary specifications before executing the configure script. For example: CFLAGS="-O `getconf LFS_CFLAGS`" LDFLAGS=`getconf LFS_LDFLAGS` LIBS=`getconf LFS_LIBS` If you try to build a large-file compatible netCDF library, then be sure to first execute the command "make distclean" in the top-level netCDF source directory. The above won't work under Solaris 2.6 or 2.8 because the "off_t" type was changed to be non-standard-conforming. Regards, Steve Emmerson <http://www.unidata.ucar.edu> -------- Original Message >Date: Wed, 14 Mar 2001 15:30:14 -0500 >From: Hugh Ellis <address@hidden> >Organization: Johns Hopkins University >To: address@hidden >Subject: Thanks >Keywords: 200103141703.f2EH3KL02469 > > Thanks for the quick and helpful response. I think I found what the problem > might be. The output file in question has size 2,147,483,032 bytes when > CCTM crashes. This is just at the 2Gb limit (2,147,483,648) which, for > Solaris 2.6, means problems (we've run into this before, trying to generate > multi-day MM5 simulations). There exists a "large-file" option for Solaris > 2.6, but it's sort of haphazard - at least as far as I can figure out. One > can have some compiled model components large-file-aware, and others not, > which is a real pain. > > Off-hand, is there any explicit provision within netCDF that may enable or > prohibit large-file processing? My guess is no - this is strictly a Solaris > problem. > > Thanks again, > > Hugh > > FYI - We were pretty much forced to use Solaris 2.6 as part of adopting the > Models3 framework (or so we thought). That has, unfortunately, proven to be > a very troublesome constraint. > > ################################################################# > ## Hugh Ellis ## > ## Professor and Chairman ## > ## Department of Geography and Environmental Engineering ## > ## Ames 312 ## > ## The Johns Hopkins University ## > ## Charles and 34th Streets ## > ## Baltimore, Maryland 21218 ## > ## ## > ## Tel: 410-516-6537 ## > ## Fax: 410-516-8996 ## > ## Email: address@hidden ## > #################################################################