Closed tpilz closed 3 years ago
Hi Tobias, Sorry, I was busy in May, and this comment was somehow leaked from my task list. If you can fix the issue easily, could you please send us the revised code? Then we can merge your commit to the master branch. (We don't use LINTERP=.FALSE. options usually)
Hi Dai,
I had a look. I think the problem is that the subroutine for the alternative interpolation is missing. I.e., according to the manual, if LINTERP=.FALSE.
nearest point interpolation should be performed and in an older version of camaflood I found the subroutine CONV_RESOL
which is doing that. This subroutine is not present in the latest code version anymore.
Is there a reason why it has been removed? Otherwise it could just be included again as in the old version.
The code structure is largely modified, and some changes are not reflected on the unused old schemes. If you can modify the code and share it with us, we can integrate your contribution
Hi Tobias, Your request is now merged to master. (This is the first attempt to merge commits by external contributors, indeed. Thanks!)
Hi, I get a segmentation fault when setting parameter
LINTERP=.FALSE.
. I think the reason is that subroutineCMF_INPMAT_INIT_*
is only called whenLINTERP
is set to true but subroutineROFF_INTERP
is called in any case, where it is then missing variablesINPX
andINPY
that are only initialised inCMF_INPMAT_INIT_*
.The error can be reproduced with the examples (I checked with
test3-jpn_fcast.sh
).I use version 4.01 (official release 20210331 obtained from the homepage).
Regards, Tobias