(Note startup time regression was a waived P2 defect for 6.0.)

New definition of startup to be measured: Cold start of the full IDE with some projects open including time to open the first java file via "Go to Type" dialog (reflects real user scenario to get a ready-to-use IDE)

• P3: Feature on Demand - improve startup time by 30%
• P2: Class and resource caching for startup (prefetch) - improve startup time by 30% (>50% together with FoD)
• P2: Project opening does not block the IDE startup - postpone project opening after startup, open the first project based on last selected file in editor
• P2: "Go to Type" dialog working even during the up-to-date scanning
• P3: Avoid initial JDK classes scanning when creating/opening first project.

2. Memory

• P2: Find memory leaks in common areas - to make sure basic open/edit/save/compile/debug/test scenarios in main project types do not create major leaks in key data structures; special focus on visual web and JSP editing
• P2: Regression tests to prevent memory leaks from being reintroduced
• P3: Infrastructure for watching class loading (to eliminate unnecessary classes during startup)
• P3: Feature on demand - to reduce static memory footprint (perm, heap)

3. I/O and networking

• P1: Filesystem optimization - reducing memory (number of file objects in memory), improvement of FS refresh time by 40%
• P1: Significantly faster files deleting + faster first modification of an opened file
• P3: Analyze disk access during common operations like expanding a package - to be able to optimize for network and slow filesystems
• P3: Analyze why some I/O related operations are significantly slower on Windows

4. Navigation

• P2: "Go to Type" dialog working even with incomplete results, possibly merged with "Go to File" functionality
• Switching between files in editor or changing selection in Explorer is slow (due to non-scaling listeners)
  • P2: Ensure responsive switching over multiple tabs (window system)
  • P2: Fix most offending listeners (especially noticeable in full IDE)
  • P3: Make repetitive actions in explorer (browsing, selections) fluent (actions system)

5. Editor

• Java editor:
  • P3: incremental parsing in method bodies - more fluent typing, faster editor hints and code completion
• JSP editor
  • P2: Improved typing performance - improvements in JSP parser, other tasks interrupted when the user starts to type, incremental html parser
  • P2: Specific memory optimizations (tag libs)
  • P3: Automated perf tests for important usages of the editor (html, jsp)

6. Visual Web

• P1: Memory leak fixes
• 50% improvement for a page open and a table drop
  • P1: Optimized BeanInfo lookup
  • P2: UI responsiveness fixed when loading pages
  • P3: Caching table metadata not to connect to database during page open
• P3: Trim TLD descriptions in Woodstock components
• P2: On-demand binding attribute - significantly reducing size of JSP and Java files for designed pages
• P2: 50% improvement for opening page flow editor (avoiding unnecessary page syncing)
• P1: Improved runtime performance of Woodstock components (in version 4.1.1)

7. UML

• P3: Improve opening of UML diagrams (a lot of time spent in finding in model in diagrams with many connections)
• P3: Improve performance reverse engineering a java project into the same UML project more than once
• P3: Improve response time of UML explorer (selecting multiple nodes, invoking context menu)

8. Setup for future

• Extended use of automated tests: more types of tests, improved test infrastructure
• New ways of problem detection and measuring
• More analysis & prototypes for future work
• Base for sustainable future development