Patch notes: Version 1.3.0

[dabico]

Changelog

Back-end Improvements

• Overhauled the search system:
  • Replaced the Native Query builder with a Criteria Query builder based on Specifications.
  • Page size for paginated search has changed from being a method parameter to an application property.
  • Navigation and download information are now sent back as GitHub-style headers, rather than in the response body itself.
  • Added sort parameter (API-only)
• Overhauled the export system:
  • Removed opencsv dependency, we solely rely on Jackson now.
  • Search parameters are now included in JSON and XML exports.
    [Include search parameters in the exported JSON #12]
  • Folder used for storing temporary files is created/deleted on startup/shutdown using simple hooks.
• The equals and hashCode methods of model classes were written with JPA performance in mind.
• Restored removed One-to-Many dependencies from GitRepo to GitRepoLabel and GitRepoLanguage (previously only maintained on the Many side).
• Disabled open-in-view, and added explicit @Fetch annotations for entities that maintain One-to-Many relationships.
• Simplified the dto package:
  • Removed all DTOs that can be emulated with Java built-ins.
  • Simplified fields of retained DTOs to use only Java built-ins.
• Converters are now implemented according to Spring's converter pattern.
• Added caching system:
  • Results of queries invoked by clients on page startup are saved in dedicated caches.
  • Cache contents are cleared every 6 hours via a dedicated CacheEvictJob.

Front-end Improvements

• Search parameters are no longer reset when navigating back from a search.
  [Back button keeps previous search parameters #10]
• Added icons for Shell and Smalltalk.
• Added written guide on how to update the custom icon sets used by the front-end.
• Minor stylistic changes.

Codebase Cleanup

• Removed unused utility classes and their tests.
• Logging has been relegated from explicit field declarations to a Lombok-provided annotation.
• Replaced the entire Interval class hierarchy with the more stable and robust Range, provided by Guava.
• Project Object Model (pom.xml) changes:
  • Removed any unused dependencies.
  • Versions of retained dependencies bumped.
  • Updated existing, and added more project-specific meta-information.

Miscellaneous

• Added application banner.

Future Roadmap

• Refactor the CrawlJob class, as well as all the services it uses.
• Include search parameters in CSV exports as well.
• Improve the label typeahead to use Bloodhound with a Full-Text Search index, or Apache Lucene.