forked from pool/python-fastparquet
* allow loading categoricals even if not so in the pandas metadata, when a column is dict-encodedand we only have one row-group (#863)  * apply dtype to the columns names series, even when selecting no columns (#861, 859)  * don't make strings while estimating bye column size (#858)  * handle upstream depr (#857, 856) OBS-URL: https://build.opensuse.org/package/show/devel:languages:python:numeric/python-fastparquet?expand=0&rev=48
Description
No description provided
Languages
RPM Spec
100%