Release Summary - Jun 15, 2023

PLAT-3504

UPS:

UPS compactor is taking more than 48 hrs to complete

UPS compactor was taking more than 48 hours to complete the process, and a few instances were stuck in the compaction process and were not able to stop. Based on the Jstack, it was found that, alreadySeenEvent and filterDuplicateValues operations were the places the threads were found processing.

To resolve the issue the following has been done.

1. HashMap resizing

We had noticed that considerable time was being spent on resizing the HashMap, to prevent this issue, all HashSets are initialized with the size of the collections.

2. Optimized view events

With this change, alreadySeenEvent (time and space complexity of n) and filterDuplicateValues (time and space complexity of n) will not be run when we add each event.

ENG-26177

Science:

Composite Outfit - change style config to work with external IDs

We have changed style configurations and job code to work with external IDs as it is easier for Dashboard to save external ids. During the model build, external IDs are transformed to internal IDs.

ENG-26122

Find:

Boosting Package Head Products

We have introduced a new feature that enables the boosting of package head products. Users can now prioritize specific products within a package while still being able to search for a product by article number and have it displayed above packages containing that product. This enhancement enhances the user experience by ensuring the correct order of packages and products

ENG-25579

Find:

Boosting Assortment Products

We have introduced a new feature that enables the boosting of products within assortments. With this enhancement, users can prioritize assortment members in normal searches, even when a mix of assortment and non-assortment products are present. The solution utilizes the assortmentFilter parameter with a new syntax, allowing for filtering and boosting of assortment products based on specific requirements.

PLAT-3494

Streaming Catalog:

Addition of Calculation Detail JSON Blob

We have implemented an enhancement that adds support for a calculation detail JSON blob. This includes two new REST calls on the "enrichment-calculation" endpoint:

GET "/vendor/:vendorId/calculation/:id/details": Retrieves the details of a calculation, with a default value of {}.

PATCH "/vendor/:vendorId/calculation/:id/details": Updates the details of a calculation. The request body must be a valid JSON.

Additionally, the calculation detail will now be included in the REST response of all endpoints that fetch subscriptions.

ENG-26191

Science:

Catalog Enrichment Job - HDFS Import

We have enhanced the Catalog Enrichment job to support importing from HDFS. This modification allows Science to utilize the job for demonstrating new systems like ChatGPT effectively. Users can now provide the HDFS location as a parameter to the job, enabling the import stage from the specified file.

This enhancement is now available for seamless data import into the Catalog Enrichment tables.

ENG-26057

Data Engineering:

Addition of strategy_to_rectype_map in Import Lookup to S3 Process

We have introduced a new utility that enables the retrieval and transfer of "strategy_to_rectype_map" data from PostgreSQL to an S3 location. This enhancement is implemented as part of the "import_labels_to_s3" job, which facilitates the seamless transfer of lookup information from PostgreSQL to AWS.

ENG-26252

Find:

Issues noticed in newly added feature - Package Boost Filter.

A client was facing issue when adding the new package boosts, the inner products of a package were not boosted as expected when searching for the article number. The issue has been resolved now.