I noticed that the existing Hash ValueOp doesn't match the behavior of other databases like BigQuery (ibis-project/ibis#2310) or Postgres (https://ibis-project.org/docs/user_guide/extending_elementwise_expr.html)
Hash takes any type and outputs an int64, whereas BigQuery and Postgres take string/bytes and output string/bytes
I'm wondering if these should try to be reconciled? Or maybe we could add a HashBytes operator?
df['ROW_NUM'] = df.sort_values(by=['PART_NUM']).groupby(['CLAIM_NUM']).cumcount() + 1that is equivalent to performing SQL's ROW_NUMBER PARTION BY syntax, where I can then filter to the first record in each grouping (ROW_NUMBER ==1).
http://ibis-project.org is working
this page works for me too. At the top of this page you can see the links to Getting started, Documentation, Community...
So the link to Documentation is not working for me.
And my bookmarks of ibis documentation also not accessible with 404 error (e.g. https://ibis-project.org/docs/backends/impala.html)