Skip to content

Autory v3.2.2

Consistent names when combining results

When converting results to CSV or SQLite, names such as Premium and PREMIUM now share one column. This also applies to variable names in long-format output. When appending to an existing SQLite table, its column spelling is preserved. Conflicting values for the same name are reported before output is written.

See name casing when converting results.

Clearer diagnostics

  • Python-Excel runs report how many projection nodes are being run when RunProjectionNodesFilter excludes nodes.
  • autory debug hides known secret values and limits its environment information to settings relevant to Autory.
  • Invalid RunControl named ranges produce clearer errors when they do not refer to a cell or refer to multiple locations.
  • Warnings about mixed data types in SQL query results now display correctly.

See run control and viewing run logs.

Reliability fixes

  • Failed SQL logins prompt for credentials immediately instead of first waiting through repeated connection attempts.
  • Failed workbook saves no longer leave files held open by Autory's workbook writer. This prevents memory leaks for large runs.
  • Uninstalling Autory reports when the Excel add-in could not be removed, with instructions to close Excel and retry.
  • Formula handling more closely follows Excel's calculation order for mixed operators.
  • Tables loaded by featherToTable keep consistent generated names when the model and its input files are moved together within the same folder structure.
  • Shared JSON result files containing non-English characters are read and written consistently.

Saved copies of binary model workbooks

When using the VBA engine with an .xlsb model, the working snapshot in the .runlog folder is now saved as .xlsm. Your original model file is unchanged. This had a side effect: It can change the generated names of data tables read from the model workbook itself. Converting all your model workbooks from .xlsb to .xlsm is a good idea.