On 22 December 2011 13:57, Matthew Knepley <knepley at gmail.com> wrote: > On Thu, Dec 22, 2011 at 10:32 AM, Lisandro Dalcin <dalcinl at gmail.com> > wrote: >> >> On 22 December 2011 12:19, Matthew Knepley <knepley at gmail.com> wrote: >> > On Thu, Dec 22, 2011 at 9:11 AM, Chekuri S. Choudary >> > <cchoudary at rnet-tech.com> wrote: >> >> >> >> >> >> >> >> I am building an RPM for petsc-3.2-p5. I would like the users of the >> >> RPM >> >> to be able to mention the PETSC_DIR at installation time, i.e, the RPM >> >> should be relocatable. >> >> >> >> For example, >> >> >> >> rpm ?ivh ?--prefix=$HOME/petsc >> >> >> >> export PETSC_DIR=$HOME/petsc >> >> >> >> >> >> >> >> The issue is that while building the RPM, the ?configure ?prefix=$DIR? >> >> builds the package assuming that $DIR is the final location of the >> >> files. >> >> So, the only way seems to be to decide the >> >> >> >> installation path upfront and make the RPM non relocatable. >> >> >> >> >> >> >> >> Even with PETSC_DIR and PETSC_ARCH, it looks like the path is being >> >> hardcoded, i.e, once PETSc is built, the files cannot be simply moved >> >> by >> >> changing PETSC_DIR and PETSC_ARCH. I was wondering if there is >> >> workaround >> >> for this issue or if I am missing anything. >> > >> > >> > No, we do not support relocatable binary packages. There are too many >> > things >> > to do during the build. >> > >> >> Mmm... Are you shure? Looking at the code in config/install.py, I >> would say that relocating a PETSc installation tree is just a matter >> of fixing PETSC_DIR/PETSC_ARCH in a bunch of text files in conf/ ... > > > Its in at least one C file, reg,c >
Yes, but that code only matters if PETSC_USE_DYNAMIC_LIBRARIES. -- Lisandro Dalcin --------------- CIMEC (INTEC/CONICET-UNL) Predio CONICET-Santa Fe Colectora RN 168 Km 472, Paraje El Pozo 3000 Santa Fe, Argentina Tel: +54-342-4511594 (ext 1011) Tel/Fax: +54-342-4511169
