Remove hardcoded iteration number from data shuffler
Until now, the DataShuffler replaces the * pattern with the snapshot number before passing the file name to openPMD. It's better to set %T here so openPMD knows the structure of the filename. With this fix, the output data can be accessed as one data series with openPMD tooling, e.g.
> openpmd-ls Be_shuffled%T.in.h5
openPMD series: Be_shuffled%T.in
openPMD standard: 1.1.0
openPMD extensions: 0
data author: ...
data created: 2024-03-11 11:31:55 +0100
data backend: HDF5
generating machine: unknown
generating software: MALA (version: 1.2.1)
generating software dependencies: unknown
number of iterations: 2 (fileBased)
all iterations: 0 1
number of meshes: 1
all meshes:
Bispectrum
number of particle species: 0
franzpoeschel:~/git-repos/mala/examples/advanced
> openpmd-ls Be_shuffled%T.out.h5
openPMD series: Be_shuffled%T.out
openPMD standard: 1.1.0
openPMD extensions: 0
data author: ...
data created: 2024-03-11 11:31:55 +0100
data backend: HDF5
generating machine: unknown
generating software: MALA (version: 1.2.1)
generating software dependencies: unknown
number of iterations: 2 (fileBased)
all iterations: 0 1
number of meshes: 1
all meshes:
LDOS
number of particle species: 0
The Be_snapshot files in the test-data repository are affected by the same issue:
> openpmd-ls Be_snapshot%T.out.h5
openPMD series: Be_snapshot%T.out
openPMD standard: 1.1.0
openPMD extensions: 0
data author: ...
data created: 2023-05-23 15:13:58 +0200
data backend: HDF5
generating machine: unknown
generating software: MALA (version: 1.1.0)
generating software dependencies: unknown
number of iterations: 4 (groupBased)
all iterations: An error occurred while opening the specified openPMD series!
Internal error: Group/Variable-based encoding: Parse preference must be set.
This is a bug. Please report at ' https://github.com/openPMD/openPMD-api/issues'.
> openpmd-ls Be_snapshot%T.in.h5
openPMD series: Be_snapshot%T.in
openPMD standard: 1.1.0
openPMD extensions: 0
data author: ...
data created: 2023-05-23 15:13:45 +0200
data backend: HDF5
generating machine: unknown
generating software: MALA (version: 1.1.0)
generating software dependencies: unknown
number of iterations: 4 (groupBased)
all iterations: An error occurred while opening the specified openPMD series!
Internal error: Group/Variable-based encoding: Parse preference must be set.
This is a bug. Please report at ' https://github.com/openPMD/openPMD-api/issues'.
(Yes, openPMD seems to have insufficient error handling here, will need to fix separately.) EDIT: This seems to be fixed already on our dev heh
To fix this, can someone tell me how these files are created in the first place?
I just saw this PR, do you still need input here?
I just saw this PR, do you still need input here?
This is now ready for review
Looks good to me, thank you!