PetscViewerHDF5HasAttribute: if dataset doesn't exist, return immediately

Merged petsc/petsc into master

Merged petsc/petsc into master

Merge branch 'master' into knepley/fix-plex-partition-parallel

* master: (414 commits)
very small updates to users manual
syntax is define(XXX) in examples tests not defined(XXX)
Increase patc

Merge branch 'master' into knepley/fix-plex-partition-parallel

* master: (414 commits)
very small updates to users manual
syntax is define(XXX) in examples tests not defined(XXX)
Increase patchlevel to 3.10.3
petsc4py: updat to latest maint snapshot with SNESTEST updates
Fix link for CUDA example in PETSc from GPUs.html page Minor fixes to installation instructions
Mat: Fixed types in Pardiso interface
Converted example comment to latex
This example crashes (for unknown reasons, probably related to the handling of extended derived types passed as arguments) with the PGI Fortran compiler
Remove outdated requires: from top of several examples
remove outdated test requires: command from top of source file
Fix the documentation for -ksp_gmres_cgs_refinement_type
Commit f20c2d659b168217cb455989f28afe052d71a64c broke lex.py so the users manual was no longer being built. There reverted the changes to lex.py
Plex: Add PetscPartitioner to fort list Suggested-by: Jiaoyan Li <Jiaoyan.Li@inl.gov>
Doc: Fix spelling Suggested-by: Manuel Colera Rico <m.colera@upm.es>
Mat ex88: check in double or quad precision only
Fix error with PetscComplex when PETSc is compiled with C and used in C++
configure: fix typo endwith -> endswith
Sys ex23: Migrated to ts_max_time in yaml file as well.
AO ex5: Fixed missing CHKERRQ();
Mat test ex88: add tests for MatMultAdd and MatMultTranposeAdd
...

Conflicts:
src/dm/impls/plex/examples/tests/ex12.c

show more ...

Merge branch 'haplav/drop-support-hdf5-older-than-1.8.0' [PR #1268]

* haplav/drop-support-hdf5-older-than-1.8.0:
Drop support for HDF5 older than 1.8.0.
No attempts to take advantage of 1.8.0+ API t

Merge branch 'haplav/drop-support-hdf5-older-than-1.8.0' [PR #1268]

* haplav/drop-support-hdf5-older-than-1.8.0:
Drop support for HDF5 older than 1.8.0.
No attempts to take advantage of 1.8.0+ API to simplify codes, will be done progressively later
See also https://lists.mcs.anl.gov/pipermail/petsc-dev/2018-December/023812.html for discussion.

show more ...

strip preprocessor conditionals regarding HDF5 version

Merge branch 'master' into knepley/feature-snes-patch

Not sure if this has done the right thing because there are no tests
yet. Also, the fortran stubs break (it doesn't know what to do with
SNESCo

Merge branch 'master' into knepley/feature-snes-patch

Not sure if this has done the right thing because there are no tests
yet. Also, the fortran stubs break (it doesn't know what to do with
SNESCompositeType), and I don't know how to fix it.

show more ...

Merged petsc/petsc into master

Merge branch 'haplav/fix-petscviewerhdf5writeattribute-for-existing-attribute' [PR #1261]

* haplav/fix-petscviewerhdf5writeattribute-for-existing-attribute:
fix PetscViewerHDF5WriteAttribute for exi

Merge branch 'haplav/fix-petscviewerhdf5writeattribute-for-existing-attribute' [PR #1261]

* haplav/fix-petscviewerhdf5writeattribute-for-existing-attribute:
fix PetscViewerHDF5WriteAttribute for existing attribute

show more ...

Merge branch 'haplav/feature-mataij-load-from-hdf5' [PR #1232]

* haplav/feature-mataij-load-from-hdf5:
MatLoad() implementation for HDF5.
This PR done with cooperation with @Mari Pecha brings the HD

Merge branch 'haplav/feature-mataij-load-from-hdf5' [PR #1232]

* haplav/feature-mataij-load-from-hdf5:
MatLoad() implementation for HDF5.
This PR done with cooperation with @Mari Pecha brings the HDF5 matrix loader.
Its impact is twofold:
- It introduces a truly parallel reader where each rank reads its chunk of the matrix.
- It can directly read MATLAB's native MAT-file format (version 7.3) which is in fact an HDF5 flavor.

show more ...

PetscViewerHDF5HasAttribute: call H5Dopen2() via PetscStackCallHDF5Return.

Negative return value means failure so petsc should fail as well.

PetscViewerHDF5WriteAttribute: Create new attribute only if it doesn't exist yet!

Otherwise use H5Aopen_name().

PetscViewerHDF5HasAttribute: call H5Aexists() via PetscStackCallHDF5Return.

Negative value means failure so petsc should fail as well.

doc fix

Merged petsc/petsc into master

PetscViewerHDF5Load: add sanity check for unitsize returned by HDF5

PetscViewerHDF5Load,PetscViewerHDF5ReadSizes support MATLAB vectors.

MATLAB stores column vectors horizontally.

PetscViewerHDF5HasAttribute: fixed comment - it supports ANY attribute

PetscViewerHDF5HasObject checks whether given dataset exists.

Old PetscViewerHDF5HasObject renamed to PetscViewerHDF5HasObject_Internal.

fix leak in PetscViewerDestroy_HDF5()

fix PetscViewerHDF5ReadAttribute for attribute parent being group

impl. PetscViewerHDF5{Get,Set}AIJNames

Merge branch 'haplav/refactor-vecload-hdf5' [PR #1214]

* haplav/refactor-vecload-hdf5:
refactor VecLoad_HDF5, ISLoad_HDF5 and PetscViewerHDF5ReadSizes
preparatory PR for another one with matrix HDF5

Merge branch 'haplav/refactor-vecload-hdf5' [PR #1214]

* haplav/refactor-vecload-hdf5:
refactor VecLoad_HDF5, ISLoad_HDF5 and PetscViewerHDF5ReadSizes
preparatory PR for another one with matrix HDF5 I/O
preparation for loading real data into complex petsc and vice versa - not yet implemented but should be easier now - currently graceful fail

show more ...

initialize a few variables to 0 to prevent compiler complaints

fix 'next' test fail: Change visibility of PetscViewerHDF5Load_Internal to PETSC_EXTERN.

Since it is possibly used across multiple shared libraries. Renamed to PetscViewerHDF5Load.