May was a productive month across the platform, with meaningful improvements to how modelers handle Goal Seek failures, compare models, and work with large data tables. Rounding things out were a number of performance improvements and a solid round of fixes to keep things running smoothly.
✨ New Features
Goal Seek “On Failure” Handling
The Goal Seek formula now includes an On Failure parameter, giving modelers more control over what happens when a goal seek can’t be solved. Instead of always causing an error, you can now choose:
- Error – the existing behavior; the formula errors.
- Warn – the model continues using a fallback value, and a warning is written to the run logs.
- Ignore – the model continues silently using the fallback value.
When using Warn or Ignore, you can select a Default Value of either the Closest Iteration (the nearest value found during solving) or a Defined Value of your choosing.
🔧 Enhancements
Data Table Content Differences in Compare Models The Compare Models view now surfaces differences in data table contents, not just structure. When two tables have differing data, a “Data” field will appear indicating that differences exist — with a direct link to the Compare Tables view. For file-only tables, the file name difference is shown instead.
Support for File-Only Data Tables Greater Than 2GB File-only data tables larger than 2GB can now be used in projections without errors in the calculation engine. This unlocks the ability to perform lookups on very large datasets that were previously unsupported.
Goal Seek Error Details for Max Iterations (HPM / Debug / Cores) When a goal seek fails due to reaching the maximum number of iterations (as opposed to general non convergence), the error details now include the last two iteration values attempted. This gives modelers the information they need to fine-tune their settings and resolve convergence issues faster.
Improved Scenario File Error Messaging When a scenario file contains an issue, error messages now identify which scenario file is the source of the problem. This is especially helpful when a projection uses multiple scenario files, such as when yield curves are involved.
Multi-Scenario Projection Performance Improvement The finalization steps for projections with a high number of scenarios has been significantly optimized, reducing what was previously a multi-minute wait to a much faster experience.
🛠Fixes
- Fixed an issue where using the models dropdown from within a Model Development page would result in an error page when selecting a production model.
- Fixed an issue that unintentionally re-introduced case-sensitivity in xlsx files in the calc engine, causing parsing errors.
- When the Subset Count formula in a Rolling Value resolves to a non-positive number, the system now returns a clear, descriptive error rather than a generic unknown error.
- The system now correctly detects and reports an error at validation time when a product output grouping is configured with an array variable, preventing errors in HPM, and product groupings results that simply used the first array value in Cores. This setup is not allowed.
- The right-click context menu for virtual folders on the Projections page no longer gets cut off at the bottom of the screen when the page doesn’t scroll.
- Resolved a rare race condition that could cause product aggregate results to be inserted twice, slightly skewing HPM product aggregate results.
- The Seed Value field (for the random number generator) now renders correctly immediately after saving, without requiring a secondary edit to display properly.
- Folder dropdowns (used when creating, editing, or copying models) now scroll correctly when there are a large number of folders, making all folders accessible.
- Resolved an intermittent failure in the scenario preparation step that could cause projections to fail under certain conditions.
- Fixed an issue where projection run logs would reference worksheet names from an old scenario file version after the file had been replaced with a new one, causing false errors.
- The projection run log no longer shows “Starting Execution” multiple times for projections with many scenarios in HPM.