View Issue Details

IDProjectCategoryView StatusLast Update
0000003Inlyra StudioPerformancepublic2026-09-19 03:38
Reporterinlyradev Assigned Toinlyradev  
PriorityhighSeveritymajorReproducibilityalways
Status resolvedResolutionfixed 
Product Versionv0.9.0 
Target Versionv1.0.0Fixed in Versionv1.0.0 
Summary0000003: Selecting downstream tools connected to Excel Input causes multi-second UI delays
DescriptionWhen selecting downstream tools connected to an Excel Input node, the configuration panel can take several seconds to appear even when the workflow is not running.

Example workflow:

Excel Input → Formula → Select → SQLite Output

Selecting the Excel Input node itself is fast. However, selecting Formula, Select, SQLite Output, or other downstream nodes causes a noticeable delay before the properties/configuration panel is displayed.

The current schema-resolution path appears to contribute to the delay. When a downstream node is selected, RefreshSelectedConfiguration() calls WorkflowSchemaResolver.ResolveInputs() to determine the upstream schema needed by the configuration editor.

For CSV Input, schema resolution was specifically implemented to read only a bounded sample of up to 2,048 rows using a StreamReader, with a code comment noting that this is done so opening a properties editor remains responsive even for large CSV files.

Excel Input is also limited to a 2,048-row schema sample, but for OpenXML workbooks (.xlsx / .xlsm) ExcelInputNodeExecutor.ReadSchema() creates a new XLWorkbook from the Excel file before reading that sample. This means that resolving the schema can still require reopening/loading the workbook each time downstream schema information is requested.

Because downstream nodes recursively resolve their upstream schemas when their configuration panels are opened, repeated tool selection may repeatedly trigger Excel workbook access. With Excel files approximately 6–10 MB in size, this results in multi-second UI delays.

Expected behavior:
Switching between workflow tools should be nearly immediate. Previously resolved Excel schema information should be reused where possible and invalidated only when schema-affecting Excel Input settings change, such as file path, worksheet, header setting, starting row/column, or schema inference options.

Actual behavior:
Selecting downstream tools causes repeated multi-second pauses and makes the workflow editor feel sluggish.
Steps To Reproduce1. Configure an Excel Input node using a workbook approximately 6–10 MB in size.
2. Connect it to Formula → Select → SQLite Output.
3. Do not run the workflow.
4. Select Excel Input; observe that its configuration panel opens quickly.
5. Select Formula; observe a multi-second delay.
6. Select Select or SQLite Output; observe similar delays.
7. Switch back to Excel Input; observe that it opens quickly.
8. Repeat switching between downstream tools and note whether the delay occurs repeatedly.
Additional InformationThe relevant implementation detail is essentially:

RefreshSelectedConfiguration()
→ WorkflowSchemaResolver.ResolveInputs()
→ recursively walks upstream
→ Excel Input calls ExcelInputNodeExecutor.ReadSchema()
→ for .xlsx, constructs a new XLWorkbook(filePath)
→ samples up to 2,048 rows for schema inference.

So the likely improvement isn't simply “sample fewer Excel rows,” because it is already capped at 2,048. The bigger opportunity is avoiding repeated workbook opens/schema resolution, probably through schema caching with proper invalidation.
TagsNo tags attached.
Attached Files
image.png (15,967 bytes)   
image.png (15,967 bytes)   
Report TypeBug Report
EditionBoth
Operating SystemWindows 11

Activities

inlyradev

2026-09-19 03:38

administrator   ~0000003

Resolved repeated delays when selecting tools downstream of Excel Input by caching the inferred schema. Cached results are refreshed when relevant input settings or source-file metadata change. User confirmed the improvement; 66 relevant automated tests passed.

The initial workbook read can still take several seconds for large files. This is expected with the current ClosedXML implementation, which loads the workbook before sampling up to 2,048 rows. Subsequent tool selections reuse the cached schema. Improving initial-load performance remains a separate optimization.

Issue History

Date Modified Username Field Change
2026-09-16 23:58 inlyradev New Issue
2026-09-16 23:58 inlyradev File Added: image.png
2026-09-17 00:06 inlyradev Assigned To => inlyradev
2026-09-17 00:06 inlyradev Status new => assigned
2026-09-17 00:12 inlyradev Target Version => v1.0.0
2026-09-17 00:20 inlyradev Assigned To inlyradev =>
2026-09-17 00:20 inlyradev Status assigned => new
2026-09-17 00:50 inlyradev Priority normal => high
2026-09-17 01:32 inlyradev Description Updated
2026-09-17 01:37 inlyradev Steps to Reproduce Updated
2026-09-17 01:37 inlyradev Additional Information Updated
2026-09-17 07:07 inlyradev Assigned To => inlyradev
2026-09-17 07:07 inlyradev Status new => assigned
2026-09-19 03:38 inlyradev Status assigned => resolved
2026-09-19 03:38 inlyradev Resolution open => fixed
2026-09-19 03:38 inlyradev Fixed in Version => v1.0.0
2026-09-19 03:38 inlyradev Note Added: 0000003