File Release Notes and Changelog

pg_bulkload 2.4.0 Released.
It supports PostgreSQL 8.4 and has several improvements
for usability: parse error skipping, unique violation
reports, and new command line options.

[features]
- Works on 8.4, 8.3 and 8.2.
- Support skipping parse erros. At most MAX_ERR_CNT errors
  are ignored and continue loading.
- Supports loading 2G or more rows at once.
- Support build with pgxs (USE_PGXS).
- Report conflicted keys on unique violations.
- $PGDATA/pg_bulkload for load status file will be created automatically.
- New command line options to pg_bulkload.
  + -i, --infile=INFILE : input file (accepts relative path)
  + -o="key=value"  : additional options
- Accepts CSV files with CR+LF for linebreaks.

[bugfix]
- Fix SIGBUS erros on non-x86 CPUs.
- Fix loading to tables with functional btree index.
- Avoid WAL writes during btree index build.

[internals]
- Add regression tests.
- Improve profiling codes for performance debug.