cism / cism Goto Github PK

The DNO_RESCALE option is currently included as a line in the Fortran compiler flags for a few build scripts (e.g. hopper-gnu and hopper-pgi). This is forcing all scale factors to be = 1, which does not currently work with Glissade and leads to errors when reading in and writing out data. Steve has identified which scripts these are and will change them asap. Workaround is to confirm that, if building CISM w/o BISICLES, the DNO_RESCALE options is NOT included in the fortran CISM compiler flags line.

Missing 'output' directory causes test to fail to run until 'output' dir is present; missing 'scratch' dir populates root level of test case dir w/ .log files, etc.

Short-term work around is to 'mkdir output' and 'mkdir scratch' from within the ./tests/higher-order/ismip-hom/ directory.

The default config file name has a typo that presents the script from running.

When initializing a 40 km North American domain with an LGM Laurentide Ice Sheet, the initialization ice disappears on the first time step (including Greenland ice) when using certain versions of GCC. This issue does not seem to appear when using modern initialization ice.

configuration where the error occurs:
Ubuntu 16.04
Linux version 4.4.0-22-generic
gcc version 5.3.1 20160413 (Ubuntu 5.3.1-14ubuntu2.1)
NetCDF version : 4.4.3-1

Working configurations using the same sample file:
Ubuntu 14.04
Linux version 3.13.0-85-generic
gcc version 4.8.4
NetCDF Version: 1:4.1.3-7ubuntu2 (4.1.3-7 ???)

CentOS7
Linux version 3.10.0-229.14.1.el7.x86_64
gcc version 4.8.3 20140911 (Red Hat 4.8.3-9)
NetCDF (fortran) version: 4.4.2

vanishing_ice_sample.tar.gz

When running the serial code and there is a fatal error, the code does not actually abort. There is a call to parallel_stop, but in the parallel_slap version of this subroutine there is no stop

Hello.
I am trying to compile CISM with BISICLES using the linux-gnu-bisicles-cmake script.
Initially I get the error:
.... /BISICLES/BisiclesToGlimmer.H:16:25: error bike_driver.H: No such file or directory

If I change this line from bike_driver.H to BisiclesToGlimmer.H, I get further in the compile, but get the following errors related to bike_driver:

../BisiclesToGlimmer.cpp:29 error:'bike_driver_init' was not declared in this scope
../BisiclesToGlimmer.cpp:39 error:'bike_driver_run' was not declared in this scope
../BisiclesToGlimmer.cpp:46 error:'bike_driver_finalize' was not declared in this scope

Any help much appreciated.
Jeremy.

I noticed a problem with the release-cism2.1.02 tag. Normally I feel like deleting and recreating a tag is a bad idea, but I'm going to do this in this case, because @whlipscomb and I feel that few if any people have started using this tag, and its existence could cause significant confusion.

basal_water=4 option (ocean penetration parameterization for effective pressure) does not seem to be applied at the initial time, and when it is applied as subsequent times the effective pressure is incorrect.

The Ross ice shelf test case does not work out of the box because it requires a directory called 'output', and the git repo does not include this (initially empty) directory. I talked to Matt, who knows how to fix this. In the meantime, 'mkdir output' will create a new directory so that the case will run.

We should check whether any other test cases require empty directories.