Streaming read
Use read-only iteration when the workflow consumes rows without modifying the workbook.
Workbook performance
Measure the workbook operation that controls your pipeline
A faster library matters only when the customer workload, output contract, and measurement method remain comparable. WolfXL publishes bounded release-artifact results and the method used to produce them.
Choose the actual bottleneck
- Large bulk reads used by ingestion or validation jobs.
- Bulk writes used to generate report data.
- Targeted edits to existing templates.
- Repeated workbook operations whose memory use limits concurrency.
Keep workbook startup, Python transformation logic, storage latency, and downstream delivery separate so the result identifies the component that changed.
Keep the comparison honest
- Use the same input workbook and requested output.
- Record exact library, Python, operating-system, and hardware identities.
- Separate cold and warm runs.
- Record iteration count, timeout, cache state, and raw artifacts.
- Validate workbook output before accepting a timing result.
- Report failures and unsupported operations as boundaries rather than omitting them.
Select the right WolfXL mode
New workbook write
Use write-mode APIs when the output is created from scratch.
Existing workbook edit
Use modify mode when a supported operation must update an existing package while retaining unrelated source parts.
Use the published evidence as a starting point
The release benchmark identifies supported scenarios where WolfXL outperformed openpyxl 3.1.5. Treat those results as workload hypotheses. Reproduce the relevant scenario with your source, build, workbook, and environment before making a production capacity decision.
Benchmark one production path
The Production Workbook Pilot can define the workload, output checks, measurement policy, and rollout threshold for one pipeline.
Scope a performance pilot