Repository navigation
make test data smaller #785
Description
Activity
In the PR #771 I have started to replace the use of demography dataset by the demography_eurostat dataset (which is actually population_session renamed).
Is demography used in the unittests?
I don't remember...
I temporarily changed the name of demography, ran the tests and found that demography is only used in the
load_example_data()function and in the tutorial.However, there is still a weird problem with demography_eurostat.h5. I'll take a look.
After some tests I discovered that the size of demography_eurostat.h5 is huge because the session contains Axis objects with labels of dtype '<U'.
There is clearly a problem when dumping axes with unicode labels to an HDF5 file.can we solve this?
We need to convert unicode labels to bytes
- added a commit that references this issue
on Aug 26, 2019 - added 3 commits that reference this issue
on Oct 7, 2019
For a reason which escapes me, population_session.h5 is 3Mb (while the same data in .xlsx is 11.8Kb) and demography.* are a bit large too, making the total size of the larray package larger than I would prefer.