Skip to content

Conversation

@CKehl
Copy link
Contributor

@CKehl CKehl commented Mar 19, 2020

The lazy evaluation from Dask is still lacking robustness and some users still experience errors in some setups, which this PR attempts to resolve.

…on paleo-plankton to verify fixes and develop a proper testing procedure
CKehl and others added 21 commits March 19, 2020 18:25
…o-bewritten, but existing tests already pass
…anParcels/parcels into stabilize_lazy_evaluation_NEMOfields
…and a NetCDF dimension (should be self.dataset.dimensions[name].size)
…h individual entry in files. That means that many files are opened multiple times simultaneously, leading to memory errors.
…ld_chunksize and the acutallly used init_chunk_size
…er-defined field_chunksize for dicts of dicts
…ints ...' by also test the dask environment default (which is usually 128MiB)
CKehl added a commit to CKehl/parcels that referenced this pull request Mar 24, 2020
CKehl added a commit to CKehl/parcels that referenced this pull request Mar 24, 2020
…s arising from an error in the NetCDFFileBuffer. Also cleared commented code where required and adapted formatting to the PEP8 linter
CKehl added a commit that referenced this pull request Mar 26, 2020
@CKehl CKehl closed this Mar 26, 2020
@CKehl CKehl deleted the stabilize_lazy_evaluation_NEMOfields branch March 26, 2020 15:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants