ChangeLog

CHANGES
=======

4.35.0
------

* release(4.35.0): fixes for skeletons, better mesh chunking, cross section, more
* refactor: whitespace
* perf: recursive lod creation, more aggressive, max triangle target
* feat: allow disabling progress bar for sharded multires mesh creation
* fix: single mip sharded encoding (#223)
* fix(downsample): situation where padding can be negative
* fix: cloudvolume changed the location of compressed\_morton\_code
* fix(downsample): handle edge for sharded downsample
* fix(downsample): generalize factor
* feat(mesh): exclude certain segids from meshing (#221)
* feat(cli): Add support for .nii.gz (#217)
* fix: wrong spec
* perf: add strategy: Recreate
* perf: set more reasonable resource requests per a pod
* perf: try to ensure pods are evenly distributed
* fix: change sharded image transfer compress default to "auto"
* fix: change default compression of TransferTask from "gzip" to "auto"
* perf: rapid organelle hole filling for meshes and skeletons (#216)
* fix: possibly resolved a segfault in cross section analysis
* feat: better behaved fill\_holes (controllable threads)
* perf(multires meshes): vastly faster mesh chunking (#214)
* typo (#213)
* feat: low memory but slow version of cross section analysis (#212)
* fix(SkeletonTask): adjustment to all\_labels missed in refactor
* docs: update support string in image create

4.34.0
------

* release(4.34.0): adds support for nii, nrrd for image create
* feat(cli): support for nrrd, nii in igneous image create
* docs: reword --sharded help text for mesh xfer
* docs: fix help text for --sharded for skeleton transfers
* refactor: remove Storage import
* fix(graphene\_meshing): adapt to fastremap sig
* fix(ccl): adapt to new fastremap.remap call sig
* fix: remove in\_place from sharded downsample
* perf: reduce memory usage for segmentation
* test: add test for offset sharded downsampling
* fix(ImageShardDownsampleTask): fix z handling
* feat(SkeletonTask): add progress bar to fill holes
* refactor: remove unnecessary copy
* perf(SkeletonTask): reduced memory usage
* perf: lower memory skeletonization (#207)
* perf(SkeletonTask): reduce memory usage
* fix(SkeletonTask): prevent integer overflow
* fix(SkeletonTask): correct for binary image crop
* fix(SkeletonTask): handle clusters from biggest to least
* refactor(SkeletonTask): missed a spot renaming delta
* perf(SkeletonTask): increase LRU 10x
* perf(SkeletonTask): use mask\_except when object\_ids is specified
* refactor: rename delta to true\_delta to avoid overwriting
* fix(SkeletonTask): can't actually reuse morphology calculation
* feat(cli): support cross section repair time budget
* feat(SkeletonTask): add a time budget for repair
* feat(SkeletonTask): add toggle for expensive contact repair
* perf: cache results of hole filling
* fix(cross\_section): handle edge of volume better
* fix: use time.monotonic to avoid datetime hiccups in elapsed time
* fix: cross section was not extracted during merge
* fix: add or remove cross section depending on input
* chore: update AUTHORS, ChangeLog

4.33.0
------

* release(4.33.0): adds multi-mip sharded downsampling
* docs: explain why 3 was chosen
* redesign: change jxl effort to 3 based on TEM experiments
* feat: multi-mip sharded downsampling (#192)
* docs: missing comma
* feat(cli): allow view to find the next available port
* feat(view): easily visualize alternative codecs
* perf: try to stop clahe thread contention
* install: bump kimimaro to 5.4.0 to get huge improvements in cross section analysis
* install: bump kimimaro requirements to get fastest postprocessing

4.32.0
------

* release(4.31.1): easier install of opencv, igneous queue, better zarr, cli indirect
* install: switch opencv-python to opencv-python-headless
* fix(zarr): better compatibility for igneous image xfer zarr://
* feat(cli:view): adds --indirect flag to route viz through cloudvolume
* feat(cli): adds "igneous queue" submenu

4.31.0
------

* release(4.31.0): Adds --fill-missing to voxel counting
* feat(voxels count): Add --fill-missing flag to CLI
* feat(CountVoxelsTask): add fill\_missing, agglomerate, and timestamp
* fix: incorrect type
* docs: show that CLAHE is available

4.30.0
------

* release(4.30.0): adds CLAHE support and makes display of upper LOD meshes neater
* install: update license info, upgrade from BETA to PRODUCTION
* chore: update CHANGELOG
* fix(multires mesh): Eliminate mesh cracks between resolutions in neuroglancer (#200)
* feat(clahe): Add Contrast Limited Adaptive Histogram Equalization (#202)
* chore: remove py38 add py313 from classifiers
* fix: avoid multiprocessing deadlocks

4.29.0
------

* release(4.29.0): adds skeleton convert command
* fix: make missing labels more robust
* feat(cli): add igneous skeleton convert for making easy SWCs
* feat(skel merge): added dry\_run flag
* feat+test(ccl/cli): support floating point thresholds

4.28.3
------

* release(4.28.3): fixes CCL for arbitrary task shapes
* fix(ccl): task assignment and update execute\_helper w/ task number
* fix(ccl): rounding error in grid size estimation
* fix: assignment error for ccl label map files

4.28.2
------

* release(4.28.2): fixes volumetric sharded downsample
* fix(volumetric downsamples): bbox was not advancing at the correct incrment (#193)
* fix(ImageShardTransferTask): record timestamp in provenance

4.28.1
------

* release(4.28.1): fixes path for skeleton xfer
* fix: skeleton transfer task using mesh directory
* feat: add dry run feature to mesh creation routine

4.28.0
------

* release(4.28.0): adds more sophisticated hole filling to meshing, skeletons
* perf: slightly more efficient memory management
* perf: earlier deletion of hole data
* fix(MeshTask): sharded was broken by double upload
* fix: ensure holes are filled properly in border contact repair
* refactor+feat(skeleton): add hole filling to cross section
* refactor: more flexible cross section using LRU
* ci: drop py313 from testing for now (needs a bunch of work)
* ci: drop py38 add py313
* feat(SkeletonTask): add more sophisticated hole filling algorithm
* fix: handle edge effect from morphological closing
* docs: more info on what fill\_holes means
* feat(mesh): add hole filling
* refactor: change meshing to use non-transposed functions
* fix: whitespace

4.27.0
------

* release(4.27.0): improvements to cross section, mesh forge --labels flag
* feat(cli): add --labels to mesh forging
* fix: ensure x,y,z ranges can't be reversed
* fix: ensure bounds\_from\_mesh accepts dict return from mesh.get
* refactor: remove obsolete task\_execution.py script
* fix: interpret environment variables correctly
* refactor: use CLI interface instead of old task\_execution.py script
* fix: incorrect escaping in CLI message
* feat(cross section): don't repair volume boundary contacts
* feat(cli): add progress flag to skeleton forging
* fix: fastremap doesn't like tuple input
* fix: if all boundaries are volume bounds, don't repair contacts
* chore: update Igneous Dockerfile to py312
* chore: update igneous-pipeline to igneous\_pipeline per new PyPI std

4.26.0
------

* release(4.26.0): add igneous wait, output option for voxel cts, fixes
* feat: add additional output option for voxel counting
* test: ensure float32 works as a dtype input for sharded
* fix(shard): enable floating support for create\_sharded\_image\_info
* fix: ensure addition is performed as uint64
* fix(skeletons): test in skeleton forging for global voxel cts outdated
* feat(cli): add "igneous wait" command
* fix: set default compression for voxel counts to zstd
* Update validate\_provenance.py (#184)
* Add --pos option to igneous view (#183)
* fix: magnitude must be greater than 0
* fix: ensure start, end are integers

4.25.1
------

* release(4.25.1): bundle of fixes, including jxl support
* fix: compatibility with cloudvolume 11 for jxl
* refactor: simplify threshold logic
* fix: change jpegxl to jxl
* make mesh\_dir behavior consistent across create\_meshing\_tasks and create\_mesh\_manifest\_tasks (#178)
* fix: missing variable for encoding\_effort
* feat: add encoding\_effort for JPEG XL to image handling
* fix: set compression options correctly when "none" for downsample
* feat: skeletonize (most) autapses and improve cross section analysis (#179)
* fix(shards/image): fix situation where chunk.z == dataset.z
* fix: make igneous view work with matrix, tigerdata

4.25.0
------

* fix: don't import CloudVolume view
* chore: update authors, changelog
* release(4.25.0): zarr support, downsample method selection
* install: bump cloudvolume version
* feat: zarr volumes should display x,y,z,c,t instead of t,c,z,y,x
* fix: incorrect parameter for downsampling
* feat: igneous view supports more than precomputed
* install: ensure forward-compatibility with future cloud-volumes
* feat: add downsample method support to image xfer
* fix(cli): show threshold-gte and theshold-lte as integers in --help
* refactor: remove cloudvolume view import
* fix: wrong variable name in shard downsample

4.24.0
------

* release(4.24.0): some fixes, adds min and max pooling to downsampling
* feat: support min and max pooling for downsampling (#176)
* Add missing task creation imports to skeleton.py (#174)
* fix: account for num\_channels in downsample mip calculation

4.23.0
------

* release(4.23.0): support for cross section analysis on skels
* docs: mention ptq tool
* feat: for small numbers of skel labels, estimate bounding box from meshes
* redesign: remove support for pickled frag files
* fix: clamp huge\_bbox
* fix: don't require a queue for skeleton forging
* redesign: change memory target conflicting with num mips to a warning
* fix: make error message an f-string
* fix: don't expand zero-volume sharded downsamples
* perf: reduce number of external requests with crackle encoded cache
* fix: target shape centroid and then cluster bad points
* fix: floating point error downstream of center calculation
* fix: improve handling of cross sectional area contacts
* feat: allow skeletons to have integer attributes by setting flag
* feat: add stop\_layer to TransferTask
* feat: support stop\_layer for ImageShardTransferTask
* chore: update changelog
* refactor: add type annotations to SkeletonTask
* docs: Fix Typo in README (#170)
* docs: Update contrast eq command in README (#171)

4.22.1
------

* release(4.22.1): fixes downsampling issue
* fix: really fix num\_mips issue
* fix(downsample): check num\_mips is not none before comparison
* test: update tests to 3.12, remove 3.7
* feat: support fill\_holes feature for cross section estimation
* fix: vastly reduce cross section underestimates
* fix: handle numpy files with a '.' in the name more gracefully
* fix: add help message to unsupported formats for igneous image create
* refactor: PrecomputedSkeleton -> Skeleton
* feat: remove queue requirement for skeleton merging
* fix: don't duplicate cross\_sectional\_area attr in info file
* fix(cli): show default for fix branching
* feat(cli): add -n to execute a certain number of tasks

4.22.0
------

* release(4.22.0): adds --cross-section to skeleton forging
* feat(SkeletonTask): compute cross sectional area (#166)
* docs: add Ryan to authors (thank you!)

4.21.0
------

* chore: update setup.cfg
* feat: support --xrange,--yrange,--zrange in (#165)
* feat(downsample): use memory limit shape when doing limited downsamples
* feat: add --labels to skeleton forging
* fix: downsampling should not affect the current mips encoding
* fix: exclude png in another location from adding gzip encoding
* fix: don't allow gzip/br on top of jpeg or png encoding

4.20.3
------

* release(4.20.3): fixes contrast normalization CLI
* fix: crashing bug in CLI for contrast normalization

4.20.2
------

* release(4.20.2): fixes bug in downsample
* fix: update info file
* fix: error in refactoring
* feat: add encoding\_level to downsample

4.20.1
------

* release(4.20.1): fixes type of num\_mips
* fix: num\_mips should be interpreted as an int

4.20.0
------

* fix: no reference to sharding in the unsharded function
* feat: add memory\_target to unsharded downsample
* chore: update authors

4.19.3
------

* release(4.19.3): float res, performance enhancements, fix shard imgs
* fix: accomodate offset volumes for shard transfer
* fix: skeletonization when resolution is non-integer value (#158)
* fix: pass cache parameter to ShardedSkeletonMergeTask
* Reduce read requests when creating sharded skeletons and meshes (#156)
* fix: remove debugging statement
* perf: add direct transfer feature to shard transfer
* fix: default value for bounds.minpt
* fix: using the right variable name for shards
* fix(image/xfer): cutout for unsharded
* fix: shards do not recognize voxel offsets so bounds should be expanded
* fix: handling offsets and dest\_voxel\_offset better for shards

4.19.2
------

* fix: processing of mip levels in shard cutout xfer

4.19.1
------

* release(4.19.1): fix igneous create
* fix(cli/create): broken extension normalization

4.19.0
------

* release(4.19.0): updated image create
* feat(cli/create): support for npy, hdf5, and crackle files
* feat: load neuroglancer remotes in RGB mode
* fix: missing urllib import
* feat(create): support np.memmap better
* chore: remove outdated comments
* feat: igneous view takes one fewer click + automatic RGB rendering
* fix: memory target did not accomodate multiple channels
* refactor: whitespace in create downsampling tasks

4.18.0
------

* release(4.18.0): new igneous image create command
* feat: add igneous image create method for npy files

4.17.0
------

* release(4.17.0): can set minishard index encoding for image shard transfer
* fix: unlock compatible with no info file
* feat: allow setting minishard index encoding
* docs: update formatting of instructions

4.16.0
------

* release(4.16.0): features and bugfixes
* docs: "Tissiue ROI" to be more specific
* docs: discuss how to use the ROI computation
* feat: roi detection (#151)
* fix: make ccl auto mode compatible with LocalTaskQueue
* fix: don't retain sharded property when doing a cutout
* fix: missing import
* feat: use in-memory queue if no queue specified
* refactor: separate network call
* test: python 3.11
* perf: switch out compresso for crackle in ccl
* test: ccl connectomics (#150)
* test: add connectomics volume test for ccl
* chore: update supported python versions
* fix: compress option was inoperative for voxel sum
* docs: remove reference to a database
* fix: use layerpath instead of cloudpath
* docs: better formatting of clean command
* feat: remove temp files for unsharded
* chore: whitespace
* feat: add sharded cleaning function for meshes and skeletons

4.15.2
------

* release(4.15.2): Fixes unbound local error
* feat: reorder task (#149)
* fix unbound local error in image xfer --cutout (#148)
* fix: 3D downsampling for sharded images

4.15.0
------

* release(4.15.0): --compress in voxel sum, better caching of voxel counts
* perf: cache voxel counts file as compressed zstd in memory
* feat: add compression option to voxel counts

4.14.0
------

* release(4.14.0): voxel ct more efficient, fix sharded ds, fix encoding
* fix: match original encoding by default during xfer
* fix: downsample encoding should match previous encoding by default
* fix(downsample): border effect for sharded downsample
* fix(downsample): sharded downsample fails for some non-zero offsets
* fixtest: voxel counting test working
* fix: update content-type for intmap
* perf: switch to using IntMap and batch JSON for computing voxel sums
* fix: needed to multiply not divide for lower mips
* fix: adjust bounds iteration when cutout is specified

4.13.0
------

* release(4.13.0): many fixes and improvements since 4.12
* fix: add cutout feature to sharded image transfers
* fix: ensure mesh info doesn't get overwritten
* fix: consolidate was added to the wrong position
* perf: download voxel counts if remote
* feat: add progress and don't check crc for counts
* feat: add bounds and "cutout" option to image xfer
* feat: use global voxel counts for dust threshold (#146)
* feat: compute global voxel counts per label (#145)
* fix: need to call consolidate during sharded skel merge
* docs: update blurb about sharded on CLI
* feat: allow disabling data\_encoding=gzip for shards

4.12.0
------

* feat+perf: adds max\_labels\_per\_shard to skeleton shard production
* fix: missing parameter, adjust docs
* refactor(image): remove unused "CloudVolumePrecomputed" import
* docs: -x is now reliable for sqs
* docs: remove word "beta" from command line
* docs: update citation file to point to paper
* docs: fix link to paper
* docs: add video tutorial link
* docs: add paper badge
* fix(ccl): clean dest info for relabel

4.11.1
------

* release(4.11.1): allow --memory to work for image xfer
* fix(cli/xfer): stop overriding --memory for xfer command

4.11.0
------

* release(4.11.0): improvements and fixes to multires mesh transfers
* redesign(mesh/multires): make default nlod = 0 instead of 1
* redesign(mesh/multires): reduce min\_chunk\_size to 256 from 512
* redesign(mesh/multires): incr. aggressiveness from 5 to 5.5
* fix(multires/mesh): format was corrupted by degenerate objects
* feat(cli): add nlod to mesh xfer
* fix(mesh xfer): better degenerate detection, check zero length meshes
* fix: make it easier to transfer unsharded to sharded meshes
* fix(skeletons): better centering in meshes

4.10.0
------

* release(4.10.0): add truncate-scales flag to xfer and fixes z chunk size
* feat(image xfer): add truncate\_scales flag and parameter
* fix: handle larger z chunk size than initial

4.9.0
-----

* release(4.9.0): adds --encoding-level to image xfer
* feat: add encoding level to transfer tasks (#140)
* docs: update ChangeLog

4.8.0
-----

* release(4.8.0): -x works for SQS, don't encode degenerate meshes
* feat(cli): add message noting that sqs is waiting
* refactor(cli): use CloudPath param type instead of extra loc
* feat(cli): add CloudPath parameter class
* docs: note why the sqs empty function is complicated
* feat(sqs): make -x flag work for SQS (120 second wait before exit)
* fix: don't try to encode degenerate meshes

4.7.0
-----

* install: ensure trimesh has needed deps
* install: add mysql-connector-python to Dockerfile
* docs: fix spelling error
* refactor: change constant to appropriate variable
* perf: parallelize each shard access
* perf(skeletons): only fetch necessary skeletons from remote shard
* perf: don't download the whole remote shard for meshes
* install: update dependencies
* fix: wrong name for shard-computer dependency
* perf: much faster shard hash calculation
* feat(spatial\_index): add --allow-missing, parallel, max\_labels\_per\_shard
* build: issue with numpy 1.23.0 through 1.23.2
* build: build and push to dockerhub
* build: try pushing docker
* chore: update docker to python3.9
* Create docker-image.yml
* feat: add pyproject.toml

4.6.0
-----

* release(4.6.0): fixes several ccl issues and adds --dust argument
* ci: ensure numpy installed first
* docs: add dust\_threshold to documentation
* test: add py37 and py310 to test roster
* fix(ccl): alignment issue near boundaries
* test: check dust works
* feat(ccl): add dust threshold
* fix(ccl): thresholding operation was incorrect

4.5.5
-----

* release(4.5.5): fixes threshold and mip issues with ccl
* chore: update Changelog and AUTHORS
* fix: prior fix for mips was off by one
* fix(cli/ccl): don't crash if you don't specify a binarization threshold
* fix: relabel tasks should copy correct mip levels

4.5.4
-----

* release(4.5.4): fix ccl cli binarization thresholds
* fix(cli/ccl): convert thresholds to integers

4.5.3
-----

* release(4.5.3): lower memory CCL, fixed factor in xfer image tasks
* perf(ccl): reduce memory usage for computing relabeling
* fix: use factor arg within create\_transfer\_tasks (#138)

4.5.2
-----

* release(4.5.2): fixes typo in unsharded->sharded meshes
* fix(mesh/multires): misspelled variable

4.5.1
-----

* release(4.5.1): fixes fill missing for ccl
* fix(ccl): fill\_missing added to task creation routines
* chore: update version

4.5.0
-----

* chore: update ChangeLog
* feat: add fill\_missing to ccl
* docs: add CCL scripting section

4.4.0
-----

* release(4.4.0): CCL without sqlite requirement + binarization thresh
* perf: scale up CCL (#136)
* release(4.3.0): adds CCL capability
* feat: whole image connected components labeling (#135)
* docs: add CITATION.cff

4.2.1
-----

* release(4.2.1): fix mesh xfer, add BASIC support multires unsharded
* fix+feat: support multires meshes for mesh xfer

4.2.0
-----

* release(4.2.0): support for multi-resolution meshing
* fix: issues in unsharded multires
* fix: update when iteration finishes
* feat: add provenance info to unsharded
* feat+refactor: adds min\_chunk\_size to merge CLI interface
* feat: multiresolution meshes (#132)

4.1.3
-----

* release(4.1.3): fix DracoPy requirement
* feat: sort mesh fragments in a consistent order before combining
* fix: give sharded meshes tighter manifest bounds
* perf(meshing): use new ambidextrous zmesh to omit a conversion

4.1.2
-----

* release(4.1.2): fixes unsharded skeleton merging
* chore: update changelog
* Some inappropriate code at UnshardedSkeletonMergeTask (#130)
* fix: skeleton sharded merge not correctly passing spatial index arg

4.1.1
-----

* release(4.1.1): fixes dependency logic for DracoPy and zmesh
* chore: allow igneous to upgrade zmesh and dracopy for feature releases

4.1.0
-----

* release(4.1.0): adds min\_shards parameter
* feat: add min\_shards parameter to shard generation
* chore: update changelog
* fix(cli): normalize path for sharded skeleton merge
* refactor: correct type annotation
* feat: add --quiet option to igneous execute
* fix: add "none" to compress options

4.0.0
-----

* release(4.0.0): reorganizes CLI, fixes contrast tasks, adds to CLI
* fix(cli): missing argument and path normalization for contrast
* fix: contrast normalization task wouldn't handle non-chunk aligned tails
* docs: add contrast commands to cli
* docs: update image based commands
* feat(cli): add "igneous image contrast" menu
* BREAKING: redesign: reoganize the image CLI commands
* docs: update sharded skeleton example
* fix: draco compression level was set too low by default
* fix: set bounds for contrast normalization task

3.2.0
-----

* release(3.2.0): Adds spatial index operations for skeletons
* chore: update ChangeLog
* feat: add spatial-index command to skeletons

3.1.0
-----

* release(3.1.0): adds delete for meshes and skels, xfer for skels
* feat: add skeleton xfer (#125)
* feat: delete meshes (#124)
* refactor: remove unnecessary imports
* chore: update ChangeLog
* docs: update the MacOS reference
* docs: update obsolete reference to task\_creation.py
* docs: mention MapBuffer in related projects

3.0.0
-----

* release(3.0.0): update xfer and downsample commands to have --encoding
* chore: update ChangeLog
* redesign(cli): change --cseg and --compresso flags to --encoding (#123)

2.0.0
-----

* release(2.0.0): offsets meshes to align with neuroglancer, closes faces
* docs: discuss why voxel\_centered=False is set for graphene
* fix(meshes): closed faces on boundary and +0.5 voxels (#121)
* perf: streamlined manifest generation on filesystem (#119)
* docs: add comment to mesh xfer

1.0.0
-----

* release(1.0.0): Adds mesh xfer command
* feat: transfer unsharded meshes to sharded meshes (#116)
* fix: save mip level of meshing during transfer
* feat: adds igenous mesh xfer to CLI
* chore: credentials.tar.enc is no longer used

0.9.3
-----

* fixtest: accomodate additional generated mip levels
* release(0.9.3): fixed a more general version of vol < chunk size
* fix: handle vol size < chunk\_size more generally

0.9.2
-----

* release(0.9.2): fixes errors caused by specifying bounds
* docs: clarify which mip level to use for xrange,yrange,zrange
* fix: each reference to get\_bounds should have a bounds\_mip argument
* fix: unused variables without definition

0.9.1
-----

* release(0.9.1): can downsample when vol size < chunk size
* fix: generate downsamples where the volume size < chunk\_size
* wip: unsharded to sharded transfer
* feat: transfer unsharded meshes to sharded meshes
* docs: examples of igneous execute -x
* docs: acks, minor bugs, mention DracoPy as a related project

0.9.0
-----

* release(0.9.0): sharding upgrades, mesh spatial index gen, CLI updates
* feat: sharded meshes (single res using trivial multi-res) (#112)
* chore: add test-suite workflow (#115)
* chore: add trimesh, bump cloud-volume in requirements
* (fix): Add kwarg for updated get\_bounds (#113)
* refactor: moved compute\_bounds script to CLI
* feat(cli): adds "igneous design bounds" command
* refactor: remove dead logger code
* fix: missing declaration for reduce
* fix: missed change of function name in refactoring
* feat: bare-bones unsharded single-res multi-resolution meshes (#111)
* (feature): Add bounds\_mip to get\_bounds (#108)
* feat(xfer): add no\_src\_update flag
* fix: don't try to write provenance files to http
* docs: add description of broken test
* fixtest: use the right content encoding
* test: check that broken dataset sums to 20 bits
* chore: update changelog

0.8.2
-----

* release(0.8.2): fixes bug in computation of shard bits
* fix: rounding errors in computing shard bits (#107)
* chore: update CHANGELOG

0.8.1
-----

* release(0.8.1): fixes bug in sharded transfers and downsamples
* fix(transfer): sharding shape computation did not account for minishard bits

0.8.0
-----

* release(0.8.0): adds --clean-info flag
* fix: QuantizeTask was broken
* feat(xfer): add --clean-info which removes skeleton and mesh metadata
* docs: update how to use zrange

0.7.0
-----

* release(0.7.0): add xrange, yrange, and zrange to downsample
* chore: delete travis
* fix(cli): pass bounds the sharded downsampling
* feat: add bounds to sharded downsamples
* feat(cli): adds xrange, yrange, and zrange to downsample

0.6.0
-----

* release(0.6.0): adds igneous downsample --sharded
* chore: remove travis ci badge (travis no longer does free open source builds)
* docs: add igneous pypi badge
* feat: shard downsampling (#106)
* refactor: change TouchTask and QuantizeTask to use queuable
* feat: add sparse support to image downsampling

0.5.0
-----

* release(0.5.0): Adds --sharded to xfer task
* chore: remove unused environment variables
* docs: info on sharded scripting and delete CLI
* chore: delete obsolete dockerfile
* feat: create image shards (#105)

0.4.1
-----

* fix(cli): --cseg was not working

0.2.0
-----

* release(0.4.0): updates to xfer and adds design command
* feat(TransferTask): fix bugs, increase performance, CLI updates (#101)
* refactor(DeleteTask): change to use queuable functions
* chore: fix deployment of README
* refactor: move infrequently used tasks to obsolete.py

0.1.0
-----

* refactor: remove [deflate] extra (it's standard nowadays)
* feat: setting up pypi package
* fix: mesh stealing bug (#96)
* chore: dracopy version number in wrong position
* chore: bump cloud-volume to match floating resolution release
* feat: support floating point resolutions (#91)
* feat(cli): add "rm image" command for deleting images
* docs: add skeleton forge and merge to README top
* docs: add "comma delimited" to CLI error message
* Implement param type for cli tuple options (#94)
* docs: describe CLI better
* chore: whitespace
* chore: bump cli version to 0.3.0
* feat(cli): add sharded merging to the skeleton command set
* feat: add unsharded skeleton merging
* feat: add max\_cable\_length arg to unsharded skeletons
* feat(cli): add skeleton forging
* feat: add fill\_holes to skeletonization options
* docs: make it easier to understand how to use the CLI
* docs: make mesh comment clearer
* feat(cli:mesh): adds mesh commands
* docs: realized we never say what Igneous does in the intro
* docs: make sure people understand the CLI tool isn't all of igneous
* docs: show how to use the CLI
* docs: describe xfer tasks better
* feat: add CLI with downsample, transfer, and execute functions
* chore: updated license Trove identifiiers
* feat: add sparse downsamples to TransferTask
* chore(MeshTask): switch from pickle to mapbuffer
* refactor: remove unnecessary imports
* fix(DeleteTask): do not redirect delete tasks
* merge: new max\_cable\_length changes
* feat(create\_sharded\_skeleton\_merge\_tasks): adds max\_cable\_length
* fix: syntax error
* feat: add max\_cable\_length to task creation
* refactor: TransferTask and DownsampleTask are queueable functions
* fix: missing z\_bits from graphene manifest generation
* fix: factor was not preserved in create\_downsample\_tasks
* fix: finally update /secrets to /root/.cloudvolume/secrets
* fix: update deployment.yaml for K8s 1.16+
* fix(TransferTask): skip\_downsamples was not being respected for scales
* feat: add max\_cable\_length argument to shardedskeletonmerge
* perf: use mapbuffer validate b/c it's faster
* perf: use mmap for sharded skeleton merge when possible (#82)
* perf: why even spend the time checking label in frag
* perf: start using MapBuffer for sharded skeletons
* refactor: change BlackoutTask to use queueable functions
* fix: wrong variable name
* perf: another attempt at reducing memory usage
* perf: attempt to reduce memory usage for ShardedSkeletonMerge
* feat: add time argument to --loop (removes --no-loop)
* chore: make deflate an optional dependency
* fix: add total argument to create progress bar for .labels uploads
* fix: don't jsonify twice
* fix: ensure correct mip is used
* feat: add performance and progress to sharded skeleton generation
* perf: faster gzip decompression
* chore: upgrade to py38
* fix: Meshing examples require tq.execute() after each phase
* docs(README): update description of supported configurations
* refactor(tasks.py): use CloudFiles instead of Storage
* refactor(skeletonization): use CloudFiles instead of Storage
* refactor: replace Storage with CloudFiles in meshing tasks
* refactor(MeshManifestTask): switch to using CloudFiles
* feat(MeshManifestTask): adds support for mesh\_dir in task\_creation
* fix: remove unnecessary requirement oauth2client
* fix: stop uploading vertext\_types with skeletons
* fix(test\_downsample\_w\_missing): failing test
* refactor: no need to maintain two implementations of chunks
* fix: add deprecated -m flag
* feat: customize the number of parallel processes
* feat: igneous now tallys completed tasks for fq
* docs: fix name of create\_skeletonizing\_tasks
* chore: bump task-queue to latest compatible version
* Allow boto3 client parameters region\_name and endpoint\_url (#75)
* docs: show how to use FileQueue
* docs: show how to use "factor" in downsample tasks
* docs: show how to use the new TaskQueue
* test: add basic skeletonization task
* chore: bump task-queue to 2.0.0
* fix: bring task\_execution.py up to date with task-queue 2.0.0
* fix: errant self reference in create\_touch\_tasks
* feat: add support for fix\_avocados in Kimimaro
* fix: syntax error when shape or dtype is omitted from decode
* fix: use variable instead of 'gzip'
* feat: MeshTask supports brotli
* fix: add module prefix to create\_downsample\_scales
* fix: inputs to add\_scale
* fix: add num\_mips to provenance file
* refactor: remove unused build function
* fix+test: downsample tasks actually use the factor argument
* feat: add factor argument to task creation for downsample and transfer
* feat: add factor argument to Downsample and Transfer
* fix: propagate fill\_missing in GrapheneMeshTask
* fix: handle compressed segmentation correctly
* feat: ensure TransferTasks can be created with compress keyword
* docs: missing table entry
* feat: support brotli compression for downsamples and transfers (#66)
* fix: make it possible to visualize skeletons while performing postprocessing
* fix: problems with fusing skeletons due to integer truncation
* fix: remove caching from ShardedSkeletonMergeTask
* fix: remove pylru as we're not using persistent caches anymore
* perf: much more efficient ShardedSkeletonMergeTask
* docs: changed name of docker to remove :master
* feat: add precalculated $shardno.labels files to sharded task generation
* fix: don't crash when there's only one segid in the whole volume
* fix: wrote sign
* fix: ensure remapping code respects 'chunks\_start\_at\_voxel\_offset'
* fix: bug in stop\_layer calculation
* chore: don't dockerize data
* fix(ShardedSkeletonMergeTask): don't error on 0 skeletons
* perf: process empty GrapheneMeshTasks faster
* feat: L2 GrapheneMeshTask (#64)
* fix: typo in upload\_batch for MeshTask
* fix: remove extraneous printing from ShardedSkeletonMergeTask
* docs: new sharded and unsharded skeletonization
* feat: sharded skeleton synthesis as an Igneous task (#61)
* fix: wrong call to cloudvolume, fix remapping
* refactor(MeshTask): use fastremap for remap and move meshdir into function
* feat(TransferTask): allow creating flat segmentations from graphene (#63)
* perf(MeshTask): speed up dust removal
* feat: add timestamp parameter to MeshTask
* feat: mesh graphene volumes
* feat(ContrastNormalizationTask): Specify lower / upper fraction separately (#62)
* feat: add synapse tracing capability (#60)
* fix: add legacy mesh type to mesh info files
* fix: set cc3d safety factor to 1
* refactor(TransferTask): translation without unaligned writes (#58)
* Make updates for cloud-volume v1.0 & remove mesh info creation (#57)
* feat: add mip and chunk\_size to mesh info file
* feat: added more sophisticated checkpoint logic
* fix: increase number of default shard bits to 9
* feat: dataset specific checkpoint names
* feat: add checkpoint to skeleton merging code
* fix: send skeleton not label to complex\_merge
* perf: reduce memory usage for single process final merging
* perf: avoid storage overhead for unfused\_skeletons
* feat: use shared memory manager for parallel processing
* feat: special handling for parallel == 1 to allow line\_profiler
* perf: reduce memory usage during unpickling shard fragments
* perf: parallel sharded skeleton generation script
* fix: strip non-float32 vertex attributes from sharded skeletons
* feat: upload mesh info file with spatial\_index chunk\_size
* feat: add spatial\_index and sharded to create\_meshing\_tasks
* feat: add support for shard fragment mesh generation
* refactor: alphabetize meshing options
* feat: add support for generating a spatial\_index for meshes
* docs: describe mesh task
* fix: some issues with 3.7 invocations
* chore: update igneous to python37
* chore: update dockerignore to reduce image sizes
* Update README.md
* feat: actually upload the shard files
* wip: checking that something is output at all
* fix: update compute\_bounds.py to use current standards
* chore: cleanup .DS\_Store
* wip: script to generate sharded skeletons
* fix: ensure that the spatial index for skeletons is non-overlapping
* fix: issue correct spatial index chunk size
* fix: incorrect variable names
* feat: support for alternative write destinations for downsampling
* refactor: Combine DownsampleTask and TransferTask
* docs: updated documentation for Igneous
* fix: correct for resolution
* fix: update spatial index field for skeleton info file
* feat: upload condensed skeleton fragments (#54)
* feat: add delete\_black\_uploads and bgcolor to downsampling tasks
* chore: CloudVolume 0.59.1 has background\_color
* chore: upgrade tqdm in requirements
* feat: TransferTasks support different background colors
* fix: "fill\_missing" was not in proveneance data for downsampling
* docs: add parameters for downsample, transfer, delete
* feat: add support for delete\_black\_uploads to TransferTask
* feat: add fill\_missing option to SkeletonTask
* perf: improve speed of object masking in mesh task
* fix: compatibility with cloudvolume 0.51.0
* refactor: use postprocessing from kimimaro 0.5.0
* fix: stable\_prefixes -> prefix\_length
* feat: add object masking to skeleton tasks
* feat: enable shared memory on kubernetes depoyments
* feat: skeleton graphene hybrid tasks
* chore: bump kimimaro version
* fix: shape and offset passed by value in FinelyDividedTaskIterator
* feat: encode created meshes in draco format (#46)
* feat: parallel argument for skeletonization
* fix: parallel should be spelled self.parallel
* feat: add parallel, progress, and dust\_threshold to skeleton options
* feat: add nano to igneous containers for editing capability
* feat: add fill\_missing to create\_meshing\_tasks
* feat: add fill\_missing option to MeshTask
* refactor: use zmesh 0.2.0 idioms
* docs: add zmesh to list of related projects
* perf: 32-bit Triangle Representations for Meshes (#49)
* Update README.md
* docs: added "Related Projects" section
* refactor: remove incorrect comment in setup.py
* fix: make will\_postprocess json serializable
* chore: specify minimum Kimimaro version
* fix: remove debug code and restore provenance for SkeletonTask creation
* feat: support Kimimaro 0.3.1's fix\_border mode (#52)
* refactor: remove skeletontricks (kimimaro already has it) (#50)
* fix: create\_delete\_tasks wasn't handling boundaries correctly
* fix: remove progress bar from production skeletonization
* perf(MeshTask): early abort if all zeros
* perf: reduce average memory pressure from meshing (#48)
* docs: update skeletonization to remove incorrect info
* fix: add chunk alignment to QuantizeTask creation
* fix: TransferTask should use return rather than yield
* feat: support for parallel upload for grid-regular tasks (#47)
* fix: missing arguments to MeshTaskIterator and new fn parameter
* fix: dust threshold was in the wrong direction..
* fix: restore mistakenly removed import of 'os'
* feat: Skeletonization (#12)
* feat: add dust\_threshold option to MeshTask
* fix: report total length as int for MeshTaskIterator and DeleteTaskIterator
* feat: skip\_downsamples flag for TransferTask
* fix: set default meshing shape to (256, 256, 256)
* fix: remove noisy progress bar from MeshManifestTasks
* fix: remove redundant progress bar from MeshTask creation
* fix: add encoding to provenance file for DownsampleTasks
* fix: use prefixes correctly in hybrid mesh manifest task
* wip: hybrid graphene mesh manifests
* chore: help protect against accidental secret uploads
* chore: ignore test2.py
* revert+fix: clamp to volume bounds in TransferTask
* fix: cap volume bounds at correct mip level for TransferTasks
* feat: introduce tinybrain (#44)
* docs: update README to reflect new TaskQueue upgrades
* fix: TransferTask boundary clamping should be handled in task creation
* fix: dvol\_bounds needed seperate definition
* fix: create\_transfer\_tasks respect destination bounds rather than source
* feat: add encoding parameter to create\_transfer\_tasks
* perf: use node DNS over cluster DNS
* fix: missing reference to LOOP
* feat: create TouchTask and BlackoutTask
* chore: bump requirements for cloudvolume and taskqueue
* refactor: remove dead code in task\_execution
* refactor: modernize task polling
* chore: make sure "projects" directory isn't included in docker images
* fix: single slice contrast normalization tasks
* fix: QueueEmpty error
* fix: integer length for DownsampleTaskIterator
* refactor: remove dead fastremap code
* chore: remove PrintTask
* refactor: Convert task\_creation.py to Iterators and TaskQueue 0.13.0 (#42)
* fix: several errors introduced by spacing and renaming
* refactor: convert all remaining tasks to iterators
* chore: remove autodeploy, it's broken
* chore: add projects to gitignore
* fix: forgot to remove some legacy attributes on ContrastNormalizationTask
* fix: missed part of refactor
* refactor: additional conversion to iterators
* wip: add DownsampleTask, DeleteTask, and MeshTask to iterator paradigm
* wip: convert task\_creation to produce iterators
* docs: update downsample documentation
* fix: get\_bounds should have only an option for expand\_to\_chunk\_size
* feat: support bounds in provenance for contrastnormalizationtask
* chore: meaningless change to force deployment
* fix: adjust downsample shape when chunk\_size is specified
* feat: add bounds to ContrastNormalizationTask
* fix: ContrastNormalizationTask respects mip level for upload
* refactor(Dockerfile): Reduce Docker image size (#37)
* fix: queue\_pickle.py can save again
* Create LICENSE
* chore: pin python-jsonschema-objects
* feat: added minval, maxval, and level\_paths to ContrastNormalizationTask
* feat: added bounds parameter to luminance levels task
* chore: added resource limits to deployment memory
* feat: add levels\_path to LuminanceLevelsTask
* chore: use --password-stdin in .travis.yml (#33)
* fix: ensure python2 installs futures
* fix: python2 compatibility
* feat: can use different encoding on downsampled layers
* fix: don't commit provenance files to BOSS
* feat: TransferTasks support transfer from any mip level
* Update README.md
* fix: write bounds to provenance file w/o crashing for TransferTask
* feat: support mesh\_dir and cdn\_cache in create\_meshing\_tasks
* feat: use $USER intead of git user.email if not available
* feat(MeshTask): configurable mesh\_dir
* changed version.py to igneous\_version.py
* feat(MeshTask): cache\_control and parallel kwargs support
* fix: broken DeleteTask
* feat: make DeleteTasks more like DownsampleTasks
* fix: abort deletion if bbox volume is zero
* feat: add provenance support for DeleteTasks
* fix: upgrade PyOpenSSL
* feat: allow deleting only some mips with DeleteTask
* fix: add bounds to transfer task provenance
* feat: chunk\_size support in create\_downsampling\_tasks
* Remove layers added by RUN (#23)
* chore: bump cloudvolume to 0.33.4
* chore: remove intern from requirements, rely on cloud-volume version
* chore: bump intern to 0.9.9
* test: honesty in testing, only use one python version
* chore: bump python to 3.6
* chore: bump python version to 3.5 from 3.4
* feat: use git user email for provenance file
* fix: reduce default mesh shape to a reasonable size
* fix: change default queue type to SQS
* Add chunkflow InferenceTask and MaskAffinitymapTask (#17)
* feat: add bounds argument to create\_transfer\_tasks
* chore: remove inaccurate intern dependency
* feat: progress bars for task upload
* chore: bump cloud-volume version
* chore: fix some style issues
* fix(MeshTask): Set default low\_padding to 0 to allow for continuous mesh boundaries
* chore: bump cloudvolume to be compatible with matrix deployment
* fix(cMesher): revert change to max\_simplification\_error from #9
* chore: update requirements.txt, CloudVolume w/ matrix support
* fix: bbox in mip 0 coordiantes
* fix: autoresizing of bounding box needed to use shape not underlying
* feat: add bounds argument, documentation, and mip=0 to create\_downsampling\_tasks
* fix(MeshTask): \_update\_vertices() accounts for 'low\_padding' offset
* fix(setup): Add numpy includes
* feat: update QuantizeAffinitiesTask to QuantizeTask
* chore: bump cloud-volume to 0.28.2
* fix: add google-cloud-logging to requirements
* chore: bump cloud-volume to 0.28.1
* feat: Cython array remap code (#14)
* fix: shape computation error in create\_deletion\_tasks
* fix: compute mip coordinates correctly in DeleteTask
* docs: documentation for sparse flag in create\_downsampling\_tasks
* refactor: move pipeline running commands to pipeline.py
* feat: add sparse downsampling capability for segmentation (#15)
* Preparations for agglomeration meshing (#9)
* refactor: simplified mesh interface, now only need to call mesher.mesh(numpy\_array) (#13)
* fix: update queue\_pickle to python3
* fix: always set chunk\_size for transfer task even if info exists
* Update README.md
* Update README.md
* Update README.md
* fix: added provenance processing for MeshTask
* fix: updated deployment.yml to reflect SQS usage
* docs: some rewording
* docs: added info on running igneous in the cloud
* docs: local multicore processing without the cloud
* chore: bump task-queue to make LocalTaskQueue available
* fix destvol dtype bug for hypersquare consensus task
* feat: automatic info file with adjustable chunk size for TransferTask
* chore: bumped python task queue to version compatible with multi-secrets
* chore: bugfix cloud-volume version
* fix: handle boundary shape correctly for ContrastNormalizationTask
* fix: handle black slices correctly in ContrastNormalizationTask
* docs: Documented downsampling, deletion, transfer, meshing, and contrast correction
* fix: ensure Python 2 can perform accurate division in ContrastNormalizationTask
* feat: ContrastNormalizationTask (#7)
* chore: upgrade cloud-volume, tqdm, and numpy
* fix: write to mip level dir for LuminanceLevelsTask
* feat: created LuminanceLevelsTask (#5)
* chore: bump cloud-volume to 0.19.1
* fix: ensure delete tasks stay within volume bounds
* feat: move provenance email to variable
* chore: bump cloudvolume to 0.15.2
* fix: added missing properties to transfer task provenance
* Improve docs a bit (#4)
* chore: bumped cloudvolume version
* chore: clean up
* fix: provenance uploads correctly for transfer tasks
* fix: cast lease\_seconds to int
* fix: added LEASE\_SECONDS to secrets.py
* feat: added ability to specify lease seconds
* chore: bumped cloudvolume to 0.10.3
* fix: fixed DeleteTask for higher level mips
* fix: HyperSquareConsensusTask working for e2198
* feat: HyperSquareConsensusTasks are now 64bit
* fix: correctly reference SQS\_URL
* chore: upgrade python-task-queue
* feat: add support for SQS
* feat: added jittered exp backoff for task\_execution
* chore: bumped cloud-volume version
* feat: added support for countless3d (#2)
* feat: marching cubes now handled uint64  (#3)
* feat: added cascade function to task\_creation
* perf: improved memory efficiency and performance of countless downsample
* feat: sanely compute the downsample chunk size when preserve\_chunk\_size=False
* chore: remove \*.pyc files from test as well
* fix: reference seunglab/igneous:master b/c that's what travis deploys
* feat: default queue is google pull-queue rather than appengine
* chore: bumped click version in requirements.txt
* chore: delete \*.pyc files
* fix: improve context swtiching on processor
* chore: bump python to 3.4 for better performance
* feat: queue\_pickly.py script can save or reload task queues
* feat: added multiprocessing to task execution
* fix: python3 compatibility for creating mesh manifest tasks
* feat: able to use appengine queue
* feat: added example deployment file for kubernetes
* chore: update python-task-queue to 0.1.7 for appengine support
* fix: push to seunglab/igneous
* fix: install in editable mode to expose \_mesher.so
* fix: include \_mesher.cpp in travis builds
* feat: added DeleteTask and PrintTask
* test: adding debug statement to see what's going on in travis
* test: updating travis file with encrypted docker information
* fix: made scripts python3 compatible
* chore: configured dependencies to enable docker build
* perf: sped up tests
* fix: python 2/3 compatibility for version.py
* fix: several installation issues + multiple travis python versions
* fix: use relative paths instead of absolute in setup.py
* fix: passing tests in python3
* docs: added Travis CI indicator to README.md
* fix: various python3 compatibility issues
* test: added credentials for travis
* init: migrated from seung-lab/neuroglancer
* Initial commit