Documented new features.
Fixed interval syntax in maf_tile specs.
Enable debugging with BIO_MAF_DEBUG environment variable.
Added debugging code.
Remove gaps when tiling.
Apparent no-op change for non_fill.
Fix interval syntax handling: chr:start-end vs. chr:start:end.
Avoid tiling output that is completely non-matching.
Only case-fold sequence data with --upcase.
Added support for a comma-separated list with --species.
For Galaxy support.
Added --fill-char and --species-file.
Refactored maf_tile to add exon concatenation feature.
Added :upcase parser option.
Correctly indicate default log destination as stderr.
Open BGZF files in binary mode.
Works around a possible bug in JRuby 1.7.0.preview2, and seems to
improve performance a bit anyway.
Updated build date.
Documented new options.
Allow setting arbitrary parser options.
Handle MAF files with no alignment blocks. Closes #112.
Improved informational logging.
Allow setting compression level.
Shut down execution on error.
Added Executor for parallel tasks in maf_bgzip.
Specifically, this lets a separate thread handle indexing under JRuby.
Log more verbosely on job completion.
Run tasks in parallel. Error if output files exist w/o --force.
Added JobRunner tool for running parallel tasks.
Uses threads on JRuby and fork() on MRI.
Only open indexes as needed in Access. Closes #109.
Previously Access would hold all the KyotoIndex objects (and
corresponding Kyoto Cabinet DB instances) open. This could lead to
problematic memory consumption with a large set of indexes.
Refactored Access to use #with_index.
This is preparatory for not holding all indexes open. It required
returning Arrays instead of Enumerables when no block is given.
Track total struct size.
Set Kyoto Cabinet options on index file.
Enable linear mode, small mode, and periodic defragmentation.
Initialize KyotoIndex properly in tests.
Validate index version. Closes #105.
Write KC changes after 1000 blocks or 50 MB.
Improved logging for maf_bgzip.
Added --one-based option. Treats all intervals as 1-based.
Needed because some tools like Ensembl output BED files using 1-based
intervals, even though this is against the standard.