uintblk expansion to 16, 32, and 64 bits

v0.8.9

add uint16blk, uint32blk, uint64blk

labelarray testing 2

Modifies the build process so dvid executables don't have an embedded web console. This will hopefully solve some Mach-O issues with Mac packaging.

labelarray testing

This release includes the new labelarray data type and compressed label block support.

• Adds first-pass support for labelarray data type that combines labelblk and labelvol
• Includes compressed label blocks via common/labels, used in some labelarray ops at this time like fast RLE generation. Internal DVID label compression is a variable compression version of the Google Neuroglancer technique. This format is only supported currently under labelarray datatype.
• Improves Google Cloud Storage support via the storage/gbucket package.

Maintenance Release

Small fixes. Added labelvol dump and annotation reload.

2017/03/20 -- Fixed bug in launch_dvid script for releases.

Maintenance update

• Fix a number of bugs and improvements to resiliency.
• Added a number of requested enhancements (see commit changes).
• Added first-pass multiscale labelblk but has not passed testing yet. Standard dvid use (i.e., not syncing labelblk to labelbk instances) will not create issues.
• First-pass float32blk support. There are plans to refactor data block datatypes to allow arbitrary dimension and # bytes per voxel.

For Linux, binaries are compiled using Go 1.7. Linux compilation uses gcc-4.8.4 due to lz4 C routines.

Groupcache and Sequencer Support

• Added "labelsz" data type, which allows ranking of labels based on number of annotations.
• Added groupcache support with configuration via TOML and HTTP API for stats.
• Added CircleCI support.
• Numerous fixes and documentation clarifications.

Binaries are compiled using Go 1.7. Linux compilation uses gcc-4.8.4 due to lz4 C routines.

Numerous Fixes

This release includes:

• A number of fixes including version ids for pushed repos, annotation syncs, kvautobus driver.
• Data instances now have a globally unique ID and can be renamed.

Multi-store Edition

This release includes:

• Refactored storage configuration that allows assignment of different stores to data types, data instances, and metadata.
• Ability to migrate data internally.
• Various fixes and improvements.

Pre-compiled packages for Linux and Mac are now available.

Hackathon Edition

Release during DVID Hackathon 2016. Changes include:

• Ability to assign different storage engines to data types and metadata.
• New DVID-to-DVID communication system employed for both command-line commands as well reworked "push" system.
• Multiple bug fixes and general cleanup

Label-aware annotations and Google cloud support

This release includes:

• Label-aware "annotations" datatype (used to be called "synapses")
• Google cloud storage engines for (1) google storage and (2) google bigtable. The google storage backend has been tested and tuned while bigtable is experimental.
• Labelvol merge caching so responses are immediate while actual merge at voxel-level is done asynchronously.
• Mutation of labelblk added versus ingest, i.e., the mutation will recreate associated sparse volumes.
• Various fixes.