Sourced from EvoTrees's releases.

v0.19.0

EvoTrees v0.19.0

Diff since v0.18.7

Breaking changes

• MLE models are not compatible with v0.18. :gaussian_mle and :logistic_mle now use Fisher information (expected Hessian) and a softplus scale instead of exp. Saved models from ≤0.18 will produce incorrect scale predictions; retrain them. User-facing offsets remain the positive scale (σ / s); internally they are mapped with invsoftplus rather than log.
• Multiclass Hessian. :mlogloss now uses the softmax diagonal p(1-p). Retraining a classifier can yield different trees; already-saved models still predict as before.
• Invalid input now errors instead of logging and continuing. In particular: a single-level classification target, a non-positive target under :gamma, an empty row subsample, and L2 / bagging_size / early_stopping_rounds outside their valid ranges.
• Classification eval labels are encoded against the training levels. y_eval must only contain classes seen in y_train; a different level order no longer silently scores the wrong prediction columns.
• MLJ update continues an existing fit only when nrounds is the sole hyper-parameter change (and is not reduced). Any other change triggers a full refit.

MLE losses: Fisher information and softplus scale

• :gaussian_mle and :logistic_mle now take Newton steps with the Fisher information (expected Hessian) rather than the observed Hessian. The Fisher matrix is diagonal and positive definite for these location-scale families, which stabilizes split finding and leaf weights.
• The unconstrained scale parameter is now softplus(x) rather than exp(x). predict still returns the positive scale (σ / s).

Multi-target regression

• Several losses can now be fit jointly on a vector of targets, with shared tree structure and a vector-valued leaf. Pass a matrix y_train of size (n_targets, nobs), or target_name=["y1", "y2", ...] on a table.
• Supported: :mse, :logloss, :poisson, :gamma, :tweedie, :mae, :quantile, :cred_std, :cred_var, :gaussian_mle, :logistic_mle. Not supported: :mlogloss, :multiquantile.
• Predictions are (nobs, n_targets). MLE losses return (nobs, 2 * n_targets) with interleaved location and positive scale per target.

GPU: KernelAbstractions, AMD and Metal

• GPU training is fully on KernelAbstractions.jl, sharing loss, metric and split-scan math with the CPU path. Oblivious trees (tree_type=:oblivious) are supported on GPU.
• New backends besides NVIDIA/CUDA: AMD/ROCm (AMDGPU.jl) and Apple Metal (Metal.jl). Load the corresponding package, then set device on the learner.
• device accepts :cpu, :gpu / :cuda (NVIDIA), :rocm / :amd, and :metal. :gpu remains an alias for CUDA.

Other

• early_stopping_tolerance: eval metric must improve by more than this value to reset the early-stopping counter (default 0.0).
• EvoTrees.predict_leaf_idx(m, data): leaf index of each observation in each tree (Matrix{UInt32} of size (nobs, ntrees)).
• Quantile eval metric now uses the model's alpha. :logistic_mle and :gini eval metrics match their intended definitions.
• importance returns zeros when a model has no splits.

Merged pull requests:

• Add multi-target regression (:mtmse): joint vector-target fitting on CPU + GPU (#328) (@​AdityaPandeyCN)
• Add early_stopping_tolerance parameter for early stopping (#331) (@​AdityaPandeyCN)
• Jdb/multi target (#333) (@​jeremiedb)
• complete kernelabstraction migration (#334) (@​jeremiedb)
• Harmonize CPU/GPU implementation (#340) (@​AdityaPandeyCN)
• Jdb/harm v2 (#342) (@​jeremiedb)
• Add predict_leaf_idx for leaf index extraction (#343) (@​AJ0070)
• Fix gini metric: relax gini_raw signature and correct argument order (#346) (@​AJ0070)
• Fix MLJ update to refit when hyperparameters other than nrounds change (#347) (@​AJ0070)
• Fix logistic_mle eval metric to match the logistic log-density (#348) (@​AJ0070)
• Error on a single-level classification target (#349) (@​AJ0070)
• Return zero importance when a model contains no splits (#351) (@​AJ0070)
• Fix llvm.prefetch on Julia 1.10 (#352) (@​AdityaPandeyCN)
• Fix poisson deviance metric for zero-valued targets (#353) (@​singhharsh1708)
• Jdb/fisher 2 (#358) (@​jeremiedb)
• Fix multiclass hessian to the softmax diagonal (#359) (@​AJ0070)
• Pass alpha to the quantile eval metric (#360) (@​AJ0070)

... (truncated)

Sourced from EvoTrees's changelog.

NEWS

v0.20.0

Model structure

• The intercept is stored on EvoTree as bias::Vector{Float32} (unconstrained / link space). trees holds only boosting rounds, so length(m.trees) == nrounds * bagging_size (and is empty when nrounds = 0).
• predict always applies bias, then the first ntree_limit trees. ntree_limit=0 is bias only.
• treeplot defaults to n=1 (the first learned tree).
• max_depth is the maximum number of splits on a leaf path (1 is a split with 2 leaves).

Plotting

• Tree plots now use a Makie recipe instead of Plots.jl / RecipesBase. Load a backend (CairoMakie or GLMakie) and call treeplot(model) or plot(model, 1).
• RecipesBase and NetworkLayout are no longer dependencies. Layout is computed in EvoTrees; the recipe is a package extension loaded with Makie.
• Split labels show the comparison (≤ for numeric/ordered features, = for unordered categoricals). Multi-output leaves show every predicted parameter.

v0.18.8

Saved :gaussian_mle / :logistic_mle models from ≤0.18.7 remain compatible: the unconstrained scale is still exp(x).

Behavioral changes

• Multiclass Hessian. :mlogloss now uses the softmax diagonal p(1-p). Retraining a classifier can yield different trees; already-saved models still predict as before.
• Invalid input now errors instead of logging and continuing. In particular: a single-level classification target, a non-positive target under :gamma, an empty row subsample, and L2 / bagging_size / early_stopping_rounds outside their valid ranges.
• Classification eval labels are encoded against the training levels. y_eval must only contain classes seen in y_train; a different level order no longer silently scores the wrong prediction columns.
• MLJ update continues an existing fit only when nrounds is the sole hyper-parameter change (and is not reduced). Any other change triggers a full refit.

MLE losses: Fisher information

• :gaussian_mle and :logistic_mle now take Newton steps with the Fisher information (expected Hessian) rather than the observed Hessian. The Fisher matrix is diagonal and positive definite for these location-scale families, which stabilizes split finding and leaf weights.
• The unconstrained scale parameter remains exp(x). predict still returns the positive scale (σ / s). User-facing offsets remain the positive scale and are mapped internally with log.

Multi-target regression

• Several losses can now be fit jointly on a vector of targets, with shared tree structure and a vector-valued leaf. Pass a matrix y_train of size (nobs, n_targets), matching x_train's (nobs, nfeats), or target_name=["y1", "y2", ...] on a table.
• Supported: :mse, :logloss, :poisson, :gamma, :tweedie, :mae, :quantile, :cred_std, :cred_var, :gaussian_mle, :logistic_mle. Not supported: :mlogloss, :multiquantile.
• Predictions are (nobs, n_targets). MLE losses return (nobs, 2 * n_targets) with interleaved location and positive scale per target.

config = EvoTreeRegressor(; loss=:mse, nrounds=200)
m = fit(config, dtrain; target_name=["y", "y2"])
pred = m(dtrain)  # size (nobs, 2)

GPU: KernelAbstractions, AMD and Metal

• GPU training is fully on KernelAbstractions.jl, sharing loss, metric and split-scan math with the CPU path. Oblivious trees (tree_type=:oblivious) are supported on GPU.
• New backends besides NVIDIA/CUDA: AMD/ROCm (AMDGPU.jl) and Apple Metal (Metal.jl). Load the corresponding package, then set device on the learner.
• device accepts :cpu, :gpu / :cuda (NVIDIA), :rocm / :amd, and :metal. :gpu remains an alias for CUDA.

Other

• early_stopping_tolerance: eval metric must improve by more than this value to reset the early-stopping counter (default 0.0).
• EvoTrees.predict_leaf_idx(m, data): leaf index of each observation in each tree (Matrix{UInt32} of size (nobs, ntrees)).
• Quantile eval metric now uses the model's alpha. :logistic_mle and :gini eval metrics match their intended definitions.
• importance returns zeros when a model has no splits.

... (truncated)

• See full diff in compare view

Sourced from DimensionalData's releases.

v0.30.2

DimensionalData v0.30.2

Diff since v0.30.1

Merged pull requests:

• Implement checkbounds with no args + with kwargs (#1183) (@​penelopeysm)
• ⚡ (Change to _noname(::AbstractString) to support Substring) (#1184) (@​arafune)
• fix indexing with sliced DimSelectors (#1190) (@​tiemvanderdeure)
• Update docs config.mts (typo) for LookupArrays link (#1192) (@​thirtysixbananas)
• Specialize Base.sizeof() for AbstractDimArray (#1196) (@​JamesWrigley)
• CompatHelper: bump compat for OrderedCollections to 2, (keep existing compat) (#1200) (@​github-actions[bot])
• Add constructor for DimArray{T,N} (#1206) (@​tiemvanderdeure)
• Makie v0.25 support (#1209) (@​icweaver)
• Make reducedims() type-stable for dim type queries (#1210) (@​JamesWrigley)
• Use optimization level 0 in the tests (#1212) (@​JamesWrigley)
• Delete unnecessary print method (#1217) (@​JamesWrigley)
• CompatHelper: bump compat for AlgebraOfGraphics in [extras] to 0.13, (keep existing compat) (#1218) (@​github-actions[bot])
• Update changelog for 0.30.2 (#1219) (@​asinghvi17)

Closed issues:

• separate set and unsafe_set so that users can't end up with broken lookups (#624)
• Make set better (#924)
• Should we start a changelog? (#983)
• checkbounds with keyword arguments (#1156)
• reducedims is not type stable (#1163)
• Indexing along a Dimension with end uses the lastindex of the entire array (#1188)
• selectors not dropped from DimSelectors after indexing (#1189)
• Broadcasting can drop a custom AbstractDimArray automatically promoting it to DimArray. (#1194)
• Custom <: AbstractDimArray can get dropped during broadcasting (#1195)
• DimArray{T,N}(undef, dims...) fails although Array{T,N}(undef, ...) works (#1202)

Sourced from DimensionalData's changelog.

[0.30.2]

Added

• DimArray{T,N} constructors for uninitialized arrays. (#1206)
• checkbounds support with no indices or dimension keyword arguments. (#1183)

Fixed

• Broadcasting arrays with SubString names and indexing sliced DimSelectors. (#1184, #1190)
• Type instability in reducedims dimension-type queries. (#1210)
• Removed a redundant print method that caused invalidations. (#1217)

Changed

• sizeof for AbstractDimArray now reports the underlying storage size. (#1196)
• Updated compatibility for AlgebraOfGraphics 0.13 and OrderedCollections 2. (#1218, #1200)
• Added forward-looking Makie 0.25 support. (#1209)

[0.30.1]

Fixed

• Restored DimensionalPlot to the main package for downstream use without RecipesBase. (#1178)

[0.30.0]

Added

• combine method added to complement groupby (#903)
• DimSlices object is now documented interface.
• Tables can now be converted to AbstractDimArray or AbstractDimStack, guessing the lookup values from dimension columns, thanks to major efforts by [@​JoshuaBillson] and [@​tiemvanderdeure] in (#739)
• presevedims can be used to preseve dimensions of array during conversion to tables, so that values may themselves be AbstractDimArray (#917)
• rebuildsliced documented and added to the developer interface
• Broadcasts improved for all AbstractBasicDimArray, like DimSelectors.
• groupby accepts name keyword to set the name of the DimGroupbyArray
• New hasinternaldimensions trait function for lookups that contain multiple dimensions - along with tests. (#991)
• BREAKING: MergedLookup and other lookups with hasinternaldimensions now return extent/bounds of their "internal" dimensions, not of their indices as they did before. If you merge an X and Y dimension, the extent of that merged lookup will have X and Y elements. (#991)

Fixed

• Begin/End bugfixed for use as types
• lookup methods like order, span etc work from all objects consistently (#1071)
• == for AbstractDimArray with missing values return missing and do not error
• isequal of AbstractDimArrays takes the dimensions into account

... (truncated)

• See full diff in compare view

Sourced from Turing's releases.

v0.49.0

Turing v0.49.0

Diff since v0.48.0

DynamicPPL 0.42.13 is now the minimum, so that SMC, PG and CSMC build a TapedTask again (#2895).

Breaking changes

NamedDist is no longer re-exported, following its removal from DynamicPPL. Replace x ~ NamedDist(dist, :y) with y ~ dist, followed by x = y if a local alias is needed (#2895).

Merged pull requests:

• Drop the NamedDist re-export and require DynamicPPL 0.42.13 (#2895) (@​shravanngoswamii)

Sourced from Turing's changelog.

0.49.0

DynamicPPL 0.42.13 is now the minimum, so that SMC, PG and CSMC build a TapedTask again (#2895).

Breaking changes

NamedDist is no longer re-exported, following its removal from DynamicPPL. Replace x ~ NamedDist(dist, :y) with y ~ dist, followed by x = y if a local alias is needed (#2895).

0.48.0

Breaking changes

The SGLD and SGHMC samplers, which did not support minibatching, have been removed along with PolynomialStepsize (#2890).

MH(var => proposal, ...) and LinkedRW have been removed. Use MH() for prior proposals, or MH(cov_matrix) for a Gaussian random walk over the complete linked parameter vector. In Gibbs, assign MH(cov_matrix) to the target variable block (#2883).

Gibbs now conditions its components with DynamicPPL.condition. GibbsContext and Turing.Inference.make_conditional are removed. Rename isgibbscomponent to supports_gibbs and gibbs_get_raw_values to gibbs_get_parameter_values; the old names are deprecated. Gibbs no longer supports model arguments containing missing: declare the latent variable inside the model and condition on observations instead (#2863).

Every variable reached by a Gibbs model must belong to a component. A component may not change the dimension or existence of a variable owned only by another component; put the variable and whatever decides its shape in one block. Changes to another block's support or distributional form remain unchecked and can make the chain reducible; see the Gibbs docstring (#2863).

Other changes

ESS Gibbs components now use the current conditional prior after another component changes its parameters (#2885).

Gibbs chains now include component-sampler statistics (#2863).

NUTS and HMCDA now accept a NamedTuple or Dict{VarName} initial_params, which HMC already did (#2878).

Emcee now honours chain_type=MCMCChains.Chains, and refuses walkers that start in different parameter layouts instead of failing inside the decode or the stretch proposal (#2879).

estimate_mode now represents indexed bounds using the model's parameter shapes, avoiding growable-array warnings during optimisation (#2888).

estimate_mode now throws on a bound that does not cover a whole variable, where it previously either dropped it and returned the unconstrained mode or failed with a bare DimensionMismatch, and warns on a bound no variable can use rather than ignoring it in silence (#2880).

Prior() now warns that initial_params has no effect instead of discarding it silently (#2881).

... (truncated)

• See full diff in compare view

Sourced from PlutoStaticHTML's releases.

v8.2.0

PlutoStaticHTML v8.2.0

Diff since v8.1.0

Merged pull requests:

• Make documenter output compatible to DocumenterCodeBlocks (#230) (@​j-fu)
• Add 1.0 to [compat] for SHA (#231) (@​j-fu)

Closed issues:

• Problems with PNG images from CairoMakie ? (#226)
• Change github pages site in repo settings ? (#228)

Sourced from PlutoStaticHTML's changelog.

[8.2.0] - 2026-09-09

Features

• Make documenter output compatible to DocumenterCodeBlocks via the documenter_code_blocks option
  • Make the identification of literal string cells configurable

[8.1.0] - 2026-08-31

Features

• Moved package to JuliaPluto
• Update docs deployment. (#225)

• Update deployment URLs to JuliaPluto
• update dependencies in example notebook
• switch cairomakie output to svg due to some errors

• Update Pluto compat to "~1.0" (#227)

• Add ExplicitImports + undocumented-names checks to the test suite
• Refine imports/precompile calls and add docstrings for OutputFormat
• Add 'pre' to CI test matrix

[8.0.0] - 2026-05-29

Features

• Update Pluto to 1.0.1 (#221)

[7.0.11] - 2026-05-20

Features

• Fix admonition header block in index.md (#216)

!!! warn not recognized by Documenter, switch to !!! warning

• Add/update syntax highlight tags to fenced code blocks (#217)
• Try ubuntu-24.04-arm for Docs.yml
• Update Pluto to 0.20.27 (#220)

[7.0.10] - 2026-02-24

Features

• Update Pluto to 0.20.23 (#215)

[7.0.9] - 2025-11-12

Features

• Switch to Pluto 0.20.21 (#213)

... (truncated)

• See full diff in compare view

Sourced from LogExpFunctions's releases.

v1.0.1

LogExpFunctions v1.0.1

Diff since v1.0.0

expand compat bounds

Merged pull requests:

• Add source path for LogExpFunctions in docs/Project.toml (#114) (@​devmotion)
• Replace CompatHelper with dependabot (#115) (@​devmotion)
• Don't add/update LogExpFunctions compat entries (#117) (@​devmotion)
• Bump the all-julia-packages group across 1 directory with 9 updates (#118) (@​dependabot[bot])
• expand version bounds for ForwardDiff (#119) (@​tpapp)

Closed issues:

• ready for 1.0.0? (#90)

Sourced from LogExpFunctions's changelog.

1.0.1

Changed

• compat bounds

1.0.0

Added

• add logabstanh

Changed

• improve log1pmx, add Float32 implementation
• make tests have their own environment, use JET.jl and Aqua.jl
• improve _log1pmx_ker internal documentation on the polynomial
• improve accuracy of logit, logcosh
• more detailed testing (ULPError.jl)
• relax the doc string of log2mexp to promise just "careful evaluation", technically breaking

0.3.29

A changelog is not available up to this version.

• fa7d47b expand compat bounds
• f06c2c1 Merge pull request #119 from JuliaStats/tp/fix-version-bounds
• 66fec94 expand version bounds for ForwardDiff
• 6df17ad Merge pull request #118 from JuliaStats/dependabot/julia/test/all-julia-packa...
• 84c0be2 Apply suggestion from @​tpapp
• b13baed Bump the all-julia-packages group across 1 directory with 9 updates
• d8cdf25 Don't add/update LogExpFunctions compat entries (#117)
• 11dd07c Replace CompatHelper with dependabot (#115)
• 6a94534 Add source path for LogExpFunctions in docs/Project.toml (#114)
• 34f13bb remove 'pre' from CI
• Additional commits viewable in compare view

Sourced from AlgebraOfGraphics's releases.

v0.13.2

AlgebraOfGraphics v0.13.2

Diff since v0.13.1

Merged pull requests:

• Minor fixes in the manuscript and bibliography (#782) (@​jbytecode)
• fix(scales): Keep verbatim data out of other layers' scales (#786) (@​jkrumbiegel)

Closed issues:

• Citing this library (#571)
• Add CONTRIBUTING.md (#778)
• verbatim and scale is required to control a second color scale (#785)

Sourced from AlgebraOfGraphics's changelog.

v0.13.2 - 2026-09-03

• Mappings with verbatim no longer pick up another layer's scale for the same aesthetic #786.

v0.13.1 - 2026-07-09

• Unitful and DynamicQuantities columns containing missing values no longer error when drawn #777.

v0.13.0 - 2026-06-25

• Analyses (linear, smooth, density, histogram, expectation, filled_contours) now fit in transformed space and back-transform their output when the relevant aesthetic carries a scale function set via scales (e.g. scales(Y = (; scale = log10))), so a fit on log-scaled data is computed in log space. This is distinct from axis = (; yscale = ...), which only transforms the display #773.
• histogram now accepts a direction (:x or :y) keyword for 1D histograms, which sets the bar orientation and the aesthetic (and thus scale space) the bins are computed in #773.
• smooth with the default degree = 2 no longer collapses on Date (and other large-magnitude) x axes #774.

v0.12.13 - 2026-06-16

• draw! now honors facet = (; size = FacetSize(...)) instead of erroring #770.

v0.12.12 - 2026-06-04

• Palette pairs and categories passed via scales now match presorted/nonnumeric-wrapped data without needing to re-wrap the keys #766.

v0.12.11 - 2026-06-02

• Fixed ABLines unit alignment rejecting a dimensionless slope when AesX and AesY carry the same unit #763.
• Multiline footnotes are now justified according to footnotealign #764.

v0.12.10 - 2026-06-02

• ABLines now aligns Unitful and DynamicQuantities units, with the intercept matching the y unit and the slope the y/x unit, and errors on incompatible units #761.

v0.12.9 - 2026-05-19

• Analyses (linear, smooth, density, histogram, expectation, frequency, contours, filled_contours) now accept Unitful.Quantity and DynamicQuantities.Quantity data #758.
• Analyses (density, histogram, linear, smooth, frequency, expectation) now drop rows where any numeric input is missing or NaN, instead of erroring or producing garbage. Inf/-Inf throw errors. #757.

v0.12.8 - 2026-05-04

• Plot types with a native :dodge attribute (BarPlot, BoxPlot, Violin, CrossBar) now also accept the generic dodge_x/dodge_y mapping, which is routed to :dodge when the aesthetic matches the plot's orientation. This makes it possible to share a single dodge_x/dodge_y mapping across companion layers (e.g., Scatter + BarPlot) without switching attribute names per layer. #747.
• Added support for Stem plots #751.

v0.12.7 - 2026-04-13

• Added FacetSize(aspect, height_fn) and facet = (; size = FacetSize(...)) for declaring an axis sizing policy that adapts to the resolved grid layout. The default wrapped() palette is now lazy and resolves to (row, col) positions at draw time using the effective axis aspect (from FacetSize or user-supplied width+height, defaulting to 1.0). For square axes this preserves the prior ceil(sqrt(n)) behavior; non-square aspects pick a column count that brings the bunched axis areas closest to a square shape #745.

v0.12.6 - 2026-03-30

• Legend entries now hide elements from plot types that have no data for a given category (hide_unused defaults to true). This fixes misleading legends when layers with different plot types share a categorical scale but cover different subsets of categories (e.g., Scatter for "A" and Lines for "B" no longer both show a marker+line for every entry). The behavior can be controlled globally via legend = (; hide_unused = false) in draw, or per scale via scales(Color = (; hide_unused_legend = false)). Paginated plots also benefit, as each page's legend now only shows categories present on that page #742.

v0.12.5 - 2026-03-23

... (truncated)

• See full diff in compare view

Sourced from DimensionalData's releases.

v0.30.2

DimensionalData v0.30.2

Diff since v0.30.1

Merged pull requests:

• Implement checkbounds with no args + with kwargs (#1183) (@​penelopeysm)
• ⚡ (Change to _noname(::AbstractString) to support Substring) (#1184) (@​arafune)
• fix indexing with sliced DimSelectors (#1190) (@​tiemvanderdeure)
• Update docs config.mts (typo) for LookupArrays link (#1192) (@​thirtysixbananas)
• Specialize Base.sizeof() for AbstractDimArray (#1196) (@​JamesWrigley)
• CompatHelper: bump compat for OrderedCollections to 2, (keep existing compat) (#1200) (@​github-actions[bot])
• Add constructor for DimArray{T,N} (#1206) (@​tiemvanderdeure)
• Makie v0.25 support (#1209) (@​icweaver)
• Make reducedims() type-stable for dim type queries (#1210) (@​JamesWrigley)
• Use optimization level 0 in the tests (#1212) (@​JamesWrigley)
• Delete unnecessary print method (#1217) (@​JamesWrigley)
• CompatHelper: bump compat for AlgebraOfGraphics in [extras] to 0.13, (keep existing compat) (#1218) (@​github-actions[bot])
• Update changelog for 0.30.2 (#1219) (@​asinghvi17)

Closed issues:

• separate set and unsafe_set so that users can't end up with broken lookups (#624)
• Make set better (#924)
• Should we start a changelog? (#983)
• checkbounds with keyword arguments (#1156)
• reducedims is not type stable (#1163)
• Indexing along a Dimension with end uses the lastindex of the entire array (#1188)
• selectors not dropped from DimSelectors after indexing (#1189)
• Broadcasting can drop a custom AbstractDimArray automatically promoting it to DimArray. (#1194)
• Custom <: AbstractDimArray can get dropped during broadcasting (#1195)
• DimArray{T,N}(undef, dims...) fails although Array{T,N}(undef, ...) works (#1202)

Sourced from DimensionalData's changelog.

[0.30.2]

Added

• DimArray{T,N} constructors for uninitialized arrays. (#1206)
• checkbounds support with no indices or dimension keyword arguments. (#1183)

Fixed

• Broadcasting arrays with SubString names and indexing sliced DimSelectors. (#1184, #1190)
• Type instability in reducedims dimension-type queries. (#1210)
• Removed a redundant print method that caused invalidations. (#1217)

Changed

• sizeof for AbstractDimArray now reports the underlying storage size. (#1196)
• Updated compatibility for AlgebraOfGraphics 0.13 and OrderedCollections 2. (#1218, #1200)
• Added forward-looking Makie 0.25 support. (#1209)

[0.30.1]

Fixed

• Restored DimensionalPlot to the main package for downstream use without RecipesBase. (#1178)

[0.30.0]

Added

• combine method added to complement groupby (#903)
• DimSlices object is now documented interface.
• Tables can now be converted to AbstractDimArray or AbstractDimStack, guessing the lookup values from dimension columns, thanks to major efforts by [@​JoshuaBillson] and [@​tiemvanderdeure] in (#739)
• presevedims can be used to preseve dimensions of array during conversion to tables, so that values may themselves be AbstractDimArray (#917)
• rebuildsliced documented and added to the developer interface
• Broadcasts improved for all AbstractBasicDimArray, like DimSelectors.
• groupby accepts name keyword to set the name of the DimGroupbyArray
• New hasinternaldimensions trait function for lookups that contain multiple dimensions - along with tests. (#991)
• BREAKING: MergedLookup and other lookups with hasinternaldimensions now return extent/bounds of their "internal" dimensions, not of their indices as they did before. If you merge an X and Y dimension, the extent of that merged lookup will have X and Y elements. (#991)

Fixed

• Begin/End bugfixed for use as types
• lookup methods like order, span etc work from all objects consistently (#1071)
• == for AbstractDimArray with missing values return missing and do not error
• isequal of AbstractDimArrays takes the dimensions into account

... (truncated)

• See full diff in compare view

Sourced from ArviZ's releases.

v0.14.2

ArviZ v0.14.2

Diff since v0.14.1

• f233782 docs: update for ArviZPythonPlots v0.2 (#435)
• 8303053 chore: Update README.md reference to CoC (#434)
• 40053eb chore: Have Code of Conduct default to org level file (#429)
• 2d884c0 chore: Bump actions/checkout from 6 to 7 (#432)
• ad37e97 chore: Bump codecov/codecov-action from 6 to 7 (#431)
• e286a41 chore: remove stale Travis-era coverage script (#430)
• 864c648 ci: replace CompatHelper with dependabot for Julia deps (#427)
• 86e8518 ci: remove invalidations workflow (#428)
• c7a9abd chore: Bump actions/github-script from 8 to 9 (#422)
• c5ebf7b chore: Bump julia-actions/setup-julia from 2 to 3 (#423)
• Additional commits viewable in compare view

Sourced from MCMCChains's releases.

v7.7.0

MCMCChains v7.7.0

Diff since v7.6.0

Remove support for PrettyTables.jl versions prior to 3.0.

Merged pull requests:

• fix prettytables version check (#500) (@​penelopeysm)

Closed issues:

• Chains internals shouldn't use (undocumented) reserved names (#469)
• Chains shouldn't (really) be indexed by Symbols (#470)

Sourced from MCMCChains's changelog.

7.7.0

Remove support for PrettyTables.jl versions prior to 3.0.

7.6.0

Compatibility for PrettyTables@3.

Minimum Julia version bumped to 1.10.

7.5.0

Add a method for MCMCDiagnosticTools.bfmi(::Chains). This computes the Bayesian Fraction of Missing Information for a chain or set of chains. Previously one had to extract a raw Array from the Chains object and pass that to bfmi.

• a95a0ee fix prettytables version check (#500)
• 9ba5d43 Add compatibility with PrettyTable v3.0 (#498)
• 7f54270 Bump version to 7.5.0, start HISTORY.md (#496)
• da56d52 BFMI function (#494)
• 2bce4e9 no coveralls, some other CI updates
• 8b5388d Prior/posterior predictive check plots (#319)
• 6909f74 Make chainsstack([c]) put info fields into vectors (#492)
• 1a03a99 Revert PrettyTables=3 compat, add a test (#490)
• de2d3ec concatenate sampler states from chains (#488)
• a45c09e CompatHelper: bump compat for PrettyTables to 3, (keep existing compat) (#485)
• Additional commits viewable in compare view

Sourced from DimensionalData's releases.

v0.30.2

DimensionalData v0.30.2

Diff since v0.30.1

Merged pull requests:

• Implement checkbounds with no args + with ...

  Description has been truncated