Healthchecks triggered in CI about hypothesis

[Zac-HD]

I'd guess the problem here is that you're imposing several interacting constraints, by picking a num_rows and num_cols up front. It's more efficient to let Hypothesis handle that internally, where it can incrementally choose whether to add each additional row. I think an efficient version of your strategy would look something like:

from hypothesis import strategies as st
from hypothesis.extra.pandas import column, data_frames, indexes

_index_strategy = st.one_of(st.text(), st.integers(), st.floats())
"""A strategy for generating dataframe indexes."""

@st.composite
def random_dataframes(draw: st.DrawFn):
    """Generate pandas dataframes of random shape and content."""
    _col_strategy = st.builds(column, name=_index_strategy, dtype=st.sampled_from([int, float, str]))
    columns = draw(st.lists(_col_strategy, max_size=5, unique_by=lambda c: c.name))
    index_strategy = indexes(elements=_index_strategy, max_size=2)
    return draw(data_frames(columns=columns, index=index_strategy))

and omitting the index_strategy (in favor of setting the first column as the index) and constraints on the number of rows would probably make this even faster.

[AdrianSosic]

Hi @Zac-HD, thanks a lot for your answer. Now that you explained the problem, it makes a lot of sense to me, and it indeed solved the issue! I was wondering: wouldn't it make sense to link something like the document you shared also in the hypothesis documentation? In fact, I was looking several times for resources about the used shrinking concepts to better understand what is happening behind the scenes (it would have probably helped my prevent the problem in the first place) but I couldn't really find much ...