Skip to content

Autory v3.5.0

VBA-free workbooks

New model workbooks and the downloadable examples now use .xlsx. Existing .xlsm and .xlsb models remain supported; you do not need to convert them to upgrade. It is recommended to convert them to .xlsx at your earliest convenience.

Patching your model workbooks with Autory v3.5.0 will remove all remaining VBA, buttons, and links. Use the Autory ribbon to run and maintain your workbooks from now on.

Improved workbook patching

There is a new Patch button on the Autory ribbon. It saves and closes the active workbook, applies all pending patches, and reopens it in the same Excel instance.

From the terminal, autory patch --max-patch N model.xlsx applies pending patches only up to and including the chosen number N.

Autory has always kept a history of patches in the Patches table on the VCS sheet. Autory now warns about missing, duplicated, out-of-order, or invalid patch records when loading models. Workbooks with no readable patch number cannot run. Patching refuses workbooks with invalid patch numbers or patches from a newer Autory version. See the patch-history guidance.

Patching is also now faster and more reliable for older models, including workbooks with copied sheets, overlapping dropdown lists, and broken named references.

Formula improvements

  • Added documentation for path(), which joins folder and file names.
  • Vector slices support arithmetic bounds, such as AVERAGE(MonthlyWage[RetAgeT-12:RetAgeT-1]).
  • Unsupported indexing syntax such as VectorA[:5] now produces an error value and a log message instead of producing a corrupt workbook when running with the Python-Excel engine.
  • Global scalar overrides now work with the Python-Julia engine.

Check installed versions

Click Versions in the ribbon's Help group, or run autory version, to see the installed Autory and Autory-Julia versions.

autory debug also includes Autory-Julia details for troubleshooting.

Other improvements

  • Filtered runs now consider dependencies beyond an excluded projection node, so nodes that match RunProjectionNodesFilter are still recalculated. Previously, excluding a node also excluded its entire dependency tree, which was unintended. See reusing previous results.
  • Source-file paths in imported data tables are clickable in Python-Excel output workbooks, making it easier to follow the audit trail.
  • In previous versions, there was a tiny risk of simultaneous runs overwriting each other's run archives and logs because of matching start times. This risk has been removed by making the hash of each run independent of the timestamp.
  • Converting results with --use-iterations is now much faster for large iteration outputs. We saw between 100x and 400x speed improvement in some models.
  • Model file extensions are recognized regardless of capitalization, including .XLSX, .JSON, and .YAML.
  • Windows commands preserve existing literal file paths instead of expanding characters in their names.
  • The documentation has a refreshed layout, with diagrams and mathematical notation no longer requiring internet access.

All new workbook patches in this release are optional. See upgrading Autory if you share models with people using older versions.