What's New

Fixed an issue where store numbers formatted as numeric values in Excel were being ignored during new product imports. The import process now correctly reads both text and numeric store values from row 2 of the spreadsheet, matching the logic already used for reading SKU values from data rows.

The import process was timing out after 30 seconds when querying for the most recent update timestamp in large Compass_IN tables. Added a composite database index on (in_store, in_db_update_datetime) that allows SQL Server to seek directly to the maximum timestamp value instead of scanning all matching rows. This eliminates the timeout and significantly improves import speed for stores with large inventory datasets.

Added a composite database index on the Compass_IN table covering both store number and update timestamp columns. This optimizes the query that determines the most recent inventory update time, which was timing out on large datasets. The existing single-column index on store number is retained for general lookups.

Fixed a crash that occurred when Mi9 CSV files contained leading or trailing whitespace in column headers. The system now properly trims and indexes column names, preventing KeyNotFoundException errors that could generate 500+ error messages in under a minute during import. The fix ensures both header validation and row-level data access use the same trimmed column names.

Fixed a KeyNotFoundException crash that occurred when Mi9 export files contained column headers with leading or trailing whitespace. The import process was already trimming headers for validation, but the row-by-row data reading was using the original untrimmed names, causing lookups to fail. The system now uses the trimmed column names consistently throughout the entire import process.

Corrected a critical flaw where imported products would default to Store='ALL' when the store number couldn't be read from the spreadsheet, causing every SKU to be fanned out across all stores in the system. The import now uses the application's primary store number as a fallback instead of 'ALL', ensuring products are always assigned to a specific store.

The Mi9 import service now clears its column index cache when processing each new file, preventing column position mismatches when importing multiple files with different structures in sequence.

The indexing logic now creates two indexes on the Compass_IN table: a single-column index on in_store for general lookups, and a composite index on (in_store, in_db_update_datetime) specifically optimized for timestamp queries during incremental sync operations.