search

ufz / ogs5

OpenGeoSys THM/C simulator version 5
https://www.opengeosys.org/ogs-5
Other
39 stars 97 forks source link

[CMake] fix switching OGS_CONFIG #81

Closed norihiro-w closed 7 years ago

norihiro-w commented 7 years ago

This PR fixes a bug reported by Luca

If cmake is run for the first time, then it sets OGS_LSOLVER to the corresponding OGS_CONFIG: if no value is given it defaults to RF. Then if the same build folder is used afterwards, with different OGS_CONFIG, the OGS_LSOLVER needs to be set manually. With cmake-gui, OGS_LSOLVER must be set manually, since it is set to RF with the first "configure". However, if OGS_LSOLVER is set manually to DEFAULT, then it is updated automatically with the right value according to OGS_CONFIG, running "configure" for the second time. Maybe there is an error and OGS_LSOLVER should be set to DEFAULT by the first "configure"? I am not very practical with cmake definitions and set-up.

In addition, another bug which might occur if one builds PETSc with external libraries is fixed.

norihiro-w commented 7 years ago

@bilke I couldn't find out why Jenkins is failing. can you please have a look?

wenqing commented 7 years ago

@norihiro-w test PETSc/THM out of the time limit.

norihiro-w commented 7 years ago

@wenqing thanks. obviously the failure is not related to this change.

wenqing commented 7 years ago

@norihiro-w I have tested on envinf.

norihiro-w commented 7 years ago

thanks. I merged this because it's a bugfix.

bilke commented 7 years ago

@norihiro-w Thanks for fixing this!