Anomalous ropp_pp installation

[Ian Culverwell]

The ROPP8.0 beta reviewer (Axel von Engeln, EUM) reported that:

  - ropp_pp installation with extra aclocal, autoconf run is a bit
odd. Can that be improved?

It's true: one often needs to restart the ropp_pp installation with

unix> aclocal -I m4 --force
unix> automake -a -c
unix> autoconf

This isn't an issue with the other modules - why?

[Ian Culverwell]

Chris Burrows reports a possible solution:

I manually compiled (with ifort12) a checked-out ROPP 
branch based on the 8.0 tag (I've been looking at 
Stig's tangent point stuff) and I had a hiccup when 
running 'make' for ropp_pp (see the output below). I 
tried running "automake --add-missing" as the error 
message suggested, and then 'make' worked fine. I don't 
think I've seen this before, so I just thought I'd let 
you know.

>make                                                    
 cd . && /bin/sh /data/local/cburrows/tp_bending/cb_tp_bending/ropp_pp/missing --run automake-1.11 --foreign --ignore-deps
configure.ac:25: warning: AC_CACHE_VAL(CM_CC_VENDOR, ...): suspicious cache-id, must contain _cv_ to be cached            
../../lib/autoconf/general.m4:1974: AC_CACHE_VAL is expanded from...                                                      
../../lib/autoconf/general.m4:1994: AC_CACHE_CHECK is expanded from...                                                    
m4/cm_compilers.m4:137: CM_PROG_CC_VENDOR is expanded from...                                                             
configure.ac:25: the top level                                                                                            
configure.ac:28: warning: AC_CACHE_VAL(CM_FC_VENDOR, ...): suspicious cache-id, must contain _cv_ to be cached            
m4/cm_compilers.m4:435: CM_PROG_FC_VENDOR is expanded from...                                                             
configure.ac:28: the top level                                                                                            
configure.ac:80: warning: AC_CACHE_VAL(ac_Module, ...): suspicious cache-id, must contain _cv_ to be cached               
m4/cm_fortran.m4:240: CM_CHECK_MODULE is expanded from...                                                                 
configure.ac:80: the top level                                                                                            
configure.ac:95: warning: AC_CACHE_VAL(ac_Module, ...): suspicious cache-id, must contain _cv_ to be cached               
configure.ac:95: the top level                                                                                            
configure.ac:108: warning: AC_CACHE_VAL(ac_Module, ...): suspicious cache-id, must contain _cv_ to be cached              
configure.ac:108: the top level                                                                                           
configure.ac:133: warning: AC_CACHE_VAL(HAVE_LIBRARY_hdf5, ...): suspicious cache-id, must contain _cv_ to be cached      
m4/cm_libraries.m4:154: CM_CHECK_LIB is expanded from...                                                                  
configure.ac:133: the top level                                                                                           
configure.ac:136: warning: AC_CACHE_VAL(HAVE_LIBRARY_netcdf, ...): suspicious cache-id, must contain _cv_ to be cached    
configure.ac:136: the top level                                                                                           
configure.ac:142: warning: AC_CACHE_VAL(HAVE_LIBRARY_ncdf, ...): suspicious cache-id, must contain _cv_ to be cached      
configure.ac:142: the top level                                                                                           
configure.ac:145: warning: AC_CACHE_VAL(HAVE_LIBRARY_ropp_io, ...): suspicious cache-id, must contain _cv_ to be cached   
configure.ac:145: the top level                                                                                           
configure.ac:146: warning: AC_CACHE_VAL(HAVE_LIBRARY_ropp_io, ...): suspicious cache-id, must contain _cv_ to be cached   
configure.ac:146: the top level                                                                                           
configure.ac:148: warning: AC_CACHE_VAL(HAVE_LIBRARY_ropp_io, ...): suspicious cache-id, must contain _cv_ to be cached   
configure.ac:148: the top level                                                                                           
configure.ac:153: warning: AC_CACHE_VAL(HAVE_LIBRARY_netcdf, ...): suspicious cache-id, must contain _cv_ to be cached    
configure.ac:153: the top level                                                                                           
configure.ac:166: warning: AC_CACHE_VAL(HAVE_LIBRARY_ropp_io, ...): suspicious cache-id, must contain _cv_ to be cached   
configure.ac:166: the top level                                                                                           
configure.ac:24: required file `./compile' not found                                                                      
configure.ac:24:   `automake --add-missing' can install `compile'                                                         
make: *** [Makefile.in] Error 1   

[Ian Culverwell]

I don't know why ropp_pp behaves in this singular way. Chris's ad-hoc solution evidently works OK. An intelligible solution probably depends on the Met Office paying a competent software engineer to sort it out.

Postponing to 'whenever', since I don't know what to do.

[Ian Culverwell]

I noticed that ropp_pp/compile was not under version control. It is generated when you run roppauto, after which ROPP compiler OK, so the absence of this file might perhaps explain why we occasionally have difficulties building ropp_pp.

The suggestion is strengthened by the realisation that the same is true of ropp_apps: no compile file, and the occasional need to run roppauto. ropp_utils also contains a compile file, but this one is under version control and is therefore downloaded.

With this in mind, ropp_pp/compile and ropp_apps/compile have been put in the trunk at r5156 and ROPP90_prototype at r5157. Perhaps it'll help.

(I noticed this when running makedistro on the trunk at r5144, which failed because compile was absent from the (new-to-the-trunk) ropp_apps module.)