search

GEOS-DEV / GEOS

GEOS Simulation Framework
GNU Lesser General Public License v2.1
210 stars 84 forks source link

Fix broken compilation on Pangea3 #3070

Closed sframba closed 5 months ago

sframba commented 5 months ago

Compilation is broken on Pangea 3 (where python component compilation is enabled), due to the following error:

src/coreComponents/common/DataTypes.hpp(83): error: the type in a dynamic_cast must be a pointer or reference to a complete class type, or void *
          detected during:
            instantiation of "NEW_TYPE geos::dynamicCast<NEW_TYPE,EXISTING_TYPE>(EXISTING_TYPE *) [with NEW_TYPE=geos::DomainPartition *, EXISTING_TYPE=geos::dataRepository::Group]" 
(99): here
            instantiation of "NEW_TYPE geos::dynamicCast<NEW_TYPE,EXISTING_TYPE>(EXISTING_TYPE &) [with NEW_TYPE=geos::DomainPartition &, EXISTING_TYPE=geos::dataRepository::Group]" 
src/coreComponents/dataRepository/Group.hpp(370): here
            instantiation of "T &geos::dataRepository::Group::getGroupByPath(const geos::string &) [with T=geos::DomainPartition]" 
src/coreComponents/physicsSolvers/python/PySolver.cpp(77): here

This is due to an include that was removed in PR 3064.

This PR reverts the needed include.

paveltomin commented 5 months ago

Sorry about that I am not sure solver base is the right place for the include, maybe place directly in pysolver? See https://fcmaelstrom.slack.com/archives/CBP94TY6A/p1712280106304179?thread_ts=1712280106.304179&cid=CBP94TY6A

codecov[bot] commented 5 months ago

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 53.20%. Comparing base (45725d5) to head (a261a6c).

Additional details and impacted files ```diff @@ Coverage Diff @@ ## develop #3070 +/- ## ======================================== Coverage 53.20% 53.20% ======================================== Files 989 989 Lines 83456 83456 ======================================== Hits 44405 44405 Misses 39051 39051 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

sframba commented 5 months ago

Sorry about that I am not sure solver base is the right place for the include, maybe place directly in pysolver? See https://fcmaelstrom.slack.com/archives/CBP94TY6A/p1712280106304179?thread_ts=1712280106.304179&cid=CBP94TY6A

I see thanks, don't worry it's really a missing CI configuration issue, we should be building pyGEOS somewhere. I moved the include to PySolver as suggested in the slack discussion