The id property of a datatable indicates the JSON key in each row, while the name is only used for display. However, many of the Canvas expressions only use the name property for finding rows. For example, here is an expression function that generates two columns with the IDs col-0-a and col-1-b and the names bytes and Sum of bytes:

esaggs index={indexPatternLoad id="90943e30-9a47-11e8-b64d-95841ca0b247"} 
  aggs={aggHistogram id="a" enabled=true schema="segment" field="bytes" maxBars=34 interval="auto" has_extended_bounds=false min_doc_count=false extended_bounds="{\"min\":\"\",\"max\":\"\"}"}
  aggs={aggSum id="b" enabled=true schema="metric" field="bytes"} metricsAtAllLevels=false partialRows=false

The only function that works as expected:

• | getCell column="col-1-b"

These functions are buggy:

• | staticColumn name="Sum of bytes" value=5: Clears all the metadata from the second columns and inserts { 'Sum of bytes': 5 } into each row, instead of updating by name.

• | columns include="col-0-a,col-1-b": Returns a totally empty table

• | staticColumn name="col-1-b" value=5: Inserts a new column instead of replacing col-1-b

• | alterColumn column="col-1-b" type="number" : Column not found

• | alterColumn column="Sum of bytes" type="number" : Inserts { 'Sum of bytes': null } into each row, which is the wrong ID.