datagrok-ai / public

Public package repository for the Datagrok.ai platform
MIT License
45 stars 26 forks source link

#2449: Calculated column type is guessed incorrectly if first 100 rows are null #2449

Open katerina-stepanova opened 11 months ago

katerina-stepanova commented 11 months ago

v1.16.7

Steps to reproduce

  1. Open SPGI
  2. Open 'Add new column' dialog > type is 'String' by default
  3. Drag 'CYP 3A Induction (PXR Reporter Gene Assay) % max' (its type is 'double') column into the formula editing area

Expected results: calculated column is auto detected as 'double' after dragging this column to the formula editor. Actual results: calculated column type is still displayed as 'String'.

image

Type is still auto detected as 'String' when this column is combined with other numerical columns using arithmetic operations:

image

Issue can be reproduced for any column where first 100 rows are null.

dnillovna commented 8 months ago

This issue has been mirrored in Jira: https://reddata.atlassian.net/browse/GROK-14616

dnillovna commented 3 months ago

This issue has been mirrored in Jira: https://reddata.atlassian.net/browse/GROK-16142