#1688: Nonexistent thorns should trigger an immediate fatal error
--------------------+-------------------------------------------------------
Reporter: hinder | Owner:
Type: defect | Status: new
Priority: minor | Milestone:
Component: Cactus | Version: development version
Keywords: |
--------------------+-------------------------------------------------------
If Cactus is built with a thornlist that contains thorns which are not
present in the Cactus tree, there should be an immediate fatal error
indicating this. Currently, the CST continues to run, building external
libraries, and finally reporting just the fact that some dependencies are
not satisfied, when the real problem is that the thorn has not been
checked out.
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1688>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1686: remove pciutils from ET list
-----------------------------------+----------------------------------------
Reporter: rhaas | Owner:
Type: defect | Status: new
Priority: blocker | Milestone: ET_2014_11
Component: EinsteinToolkit thorn | Version: development version
Keywords: |
-----------------------------------+----------------------------------------
ExternalLibraries pciutils currently does not compile on OSX (#1658), it
hard-codes the compiler to be gcc (#1657), it does not search for a system
installed version of itself (those can actually exist on Linux machines,
#1669), and mis-behaves in that its VERBOSE setting is the default (ie it
outputs all shell commands it executes in configure.sh).
libpciutils was added in ticket #1621 and was not part of the last release
so we are not bound be the decprecation policy.
Since only PAPI makes use of it, and PAPI compiles without, I would
suggest to comment out libpciutils for the release thornlist.
This is marked "blocker" due to #1658 preventing any ET build on OSX
machines.
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1686>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1661: ExternalLibraries/PAPI does not build with gcc 4.8.2
-----------------------------------+----------------------------------------
Reporter: tbode | Owner: tbode
Type: defect | Status: new
Priority: major | Milestone: ET_2014_11
Component: EinsteinToolkit thorn | Version: development version
Keywords: PAPI |
-----------------------------------+----------------------------------------
This seems to be a repeat of #1459 whose fix was never re-applied after
updating to PAPI-5.3.0. With the newer version of PAPI and newer compiler,
there are quite a few newer complaints (see attached make.log). This is
done on a machine running Ubuntu 14.04 Trusty.
I am attaching an equivalent set of patches to #1459 which seems to work
as a workaround for the moment (namely removing the -Werror flags).
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1661>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1598: ExternalLibraries/MPI does not auto-detect openmpi on Debian/testing
-----------------------------------+----------------------------------------
Reporter: rhaas | Owner:
Type: defect | Status: new
Priority: minor | Milestone:
Component: EinsteinToolkit thorn | Version: development version
Keywords: MPI |
-----------------------------------+----------------------------------------
configure.sh expects to find libmpi.a in $MPI_DIR/lib however Debian no
longer delivers a static libmpi.a but only libmpi.so so the tests fails.
The fix is likely similar to what HDF5 already does (ie accept any of .a
.so .dylib as extensions for libraries). The fix is a bit complex though,
so (given that Debian/stable is still fine) this is a reminder for after
the release. Explictly specifying MPI_DIR, MPI_INC_DIRS, MPI_LIB_DIRS and
MPI_LIBS allows this to be worked around in the current release.
For reference, the fragment from HDF5 looks like this:
{{{
# We look in these directories
DIRS="/usr /usr/local /usr/local/hdf5 /usr/local/packages/hdf5
/usr/local/apps/hdf5 /opt/local ${HOME} ${HOME}/hdf5 c:/packages/hdf5"
# look into each directory
for dir in $DIRS; do
# libraries might have different file extensions
for libext in a so dylib; do
# libraries can be in /lib or /lib64
for libdir in lib64 lib; do
# These files must exist
FILES="include/hdf5.h $(for lib in ${HDF5_CXX_LIBS}
${HDF5_FORTRAN_LIBS} ${HDF5_C_LIBS}; do echo
${libdir}/lib${lib}.${libext}; done)"
# assume this is the one and check all needed files
HDF5_DIR="$dir"
for file in $FILES; do
# discard this directory if one file was not found
if [ ! -r "$dir/$file" ]; then
unset HDF5_DIR
break
fi
done
# don't look further if all files have been found
if [ -n "$HDF5_DIR" ]; then
break
fi
done
# don't look further if all files have been found
if [ -n "$HDF5_DIR" ]; then
break
fi
done
# don't look further if all files have been found
if [ -n "$HDF5_DIR" ]; then
break
fi
done
}}}
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1598>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1684: Add PETSc and friends to Jenkins
-------------------------------------+--------------------------------------
Reporter: eschnett | Owner:
Type: enhancement | Status: new
Priority: major | Milestone:
Component: EinsteinToolkit website | Version: development version
Keywords: |
-------------------------------------+--------------------------------------
Can we add PETSc (and related thorns) to Jenkins, so that they are tested
automatically?
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1684>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1677: Support attribute((nonnull)), attribute((returns_nonnull))
-------------------------+--------------------------------------------------
Reporter: eschnett | Owner:
Type: enhancement | Status: new
Priority: major | Milestone:
Component: Other | Version: development version
Keywords: |
-------------------------+--------------------------------------------------
GCC (and probably other compilers) have attributes nonnull and
returns_nonnull. These describe functions that require either arguments or
return values to be not null, allowing both additional optimizations and
consistency checks.
I would like to add support for Cactus's autoconf mechanism to detect
this, and then use these attributes in a few central places such as
CCTK_ARGUMENTS or CCTK_PARAMETERS.
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1677>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1683: build error TATPETSc
---------------------+------------------------------------------------------
Reporter: knarf | Owner:
Type: defect | Status: new
Priority: blocker | Milestone: ET_2014_11
Component: Cactus | Version: development version
Keywords: |
---------------------+------------------------------------------------------
While building TATPETSc I get:
{{{
/home/knarf/ET_trunk/configs/openblas/build/TATPETSc/solve.c:291:23:
error: ‘MPI_UINT32_T’ undeclared (first use in this function)
/home/knarf/ET_trunk/configs/openblas/build/TATPETSc/solve.c:291:23: note:
each undeclared identifier is reported only once for each function it
appears in
/home/knarf/ET_trunk/configs/openblas/build/TATPETSc/solve.c:292:23:
error: ‘MPI_UINT64_T’ undeclared (first use in this function)
}}}
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1683>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1681: Switch SimFactory2 to git
-------------------------+--------------------------------------------------
Reporter: eschnett | Owner:
Type: enhancement | Status: new
Priority: major | Milestone:
Component: Other | Version: development version
Keywords: |
-------------------------+--------------------------------------------------
Pull request at <https://bitbucket.org/einsteintoolkit/manifest/pull-
request/2/switch-simfactory2-to-git/diff>.
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1681>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1641: Switch Slab default from "alltoall" to "irecv/isend"
-----------------------------------+----------------------------------------
Reporter: eschnett | Owner:
Type: defect | Status: new
Priority: major | Milestone: ET_2014_11
Component: EinsteinToolkit thorn | Version: development version
Keywords: |
-----------------------------------+----------------------------------------
The "alltoall" algorithm in Slab does not scale well.
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1641>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit
#1678: Update OpenSSL to 1.0.1j
-------------------------+--------------------------------------------------
Reporter: eschnett | Owner:
Type: enhancement | Status: new
Priority: major | Milestone:
Component: Other | Version: development version
Keywords: |
-------------------------+--------------------------------------------------
I have a patch that updates OpenSSL to version 1.0.1j. Ok to apply?
--
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1678>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit