# igraph C library changelog

## [master]

### Fixed

 - `igraph_community_label_propagation()` is now interruptible.
 - `igraph_is_bipartite()` would on rare occasions return invalid results when the cache was employed.
 - `igraph_weighted_adjacency()` correctly passes through NaN values with `IGRAPH_ADJ_MAX`, and correctly recognizes symmetric adjacency matrices containing NaN values with `IGRAPH_ADJ_UNDIRECTED`.
 - `igraph_read_graph_gml()` can now read GML files that use ids larger than what is representable on 32 bits, provided that igraph was configured with a 64-bit `igraph_integer_t` size.
 - Fixed a performance issue in `igraph_read_graph_graphml()` with files containing a very large number of entities, such as `>`.

### Other

 - Documentation improvements.

## [0.10.11] - 2024-04-02

### Added

 - `igraph_is_complete()` checks whether there is a connection between all pairs of vertices (experimental function, contributed by Aymeric Agon-Rambosson @aagon in #2510).
 - `igraph_join()` creates the _join_ of two graphs (experimental function, contributed by Quinn Buratynski @GanzuraTheConsumer in #2508).

### Fixed

 - Fixed a corruption of the "finally" stack in `igraph_write_graph_gml()` for certain invalid GML files.
 - Fixed a memory leak in `igraph_write_graph_lgl()` when vertex names were present but edge weights were not.
 - Fixed the handling of duplicate edge IDs in `igraph_subgraph_from_edges()`.
 - Fixed conversion of sparse matrices to dense with `igraph_sparsemat_as_matrix()` when sparse matrix object did not make use of its full allocated capacity.
 - `igraph_write_graph_ncol()` and `igraph_write_graph_lgl()` now refuse to write vertex names which would result in an invalid file that cannot be read back in.
 - `igraph_write_graph_gml()` now ignores graph attributes called `edge` or `node` with a warning. Writing these would create an invalid GML file that igraph couldn't read back.
 - `igraph_disjoint_union()` and `igraph_disjoint_union_many()` now check for overflow.
 - `igraph_read_graph_graphml()` now correctly compares attribute values with certain expected values, meaning that prefixes of valid values of `attr.type` are not accepted anymore.
 - Empty IDs are not allowed any more in `<key>` tags of GraphML files as this is a violation of the GraphML specification.
 - `igraph_is_separator()` and `igraph_is_minimal_separator()` now work correctly with disconnected graphs.
 - `igraph_linegraph()` now considers self-loops to be self-adjacent in undirected graphs, bringing consistency with how directed graphs were already handled in previous versions.
 - `igraph_all_st_mincuts()` now correctly returns all minimum cuts. This also fixes a problem with `igraph_minimum_size_separators()`.
 - Corrected minor error in `igraph_community_label_propagation()` when adding labels to isolated nodes with some fixed labels present.
 - `igraph_community_spinglass()` no longer crashes when passing an edgeless graph and an empty weight vector.
 - `igraph_rewire()` no longer crashes on graphs with more than three vertices but fewer than two edges.

### Changed

 - `igraph_rewire()` on longer throws an error on graphs with fewer than four vertices. These graphs are now returned unchanged, just like other graphs which are the unique realization of their degree sequence.

### Other

 - Performance: `igraph_is_simple()` now makes more granular use of the cache.
 - Performance: `igraph_degree()` now makes use of the cache when checking for self-loops.
 - The performance of `igraph_is_minimal_separator()` was improved.
 - `igraph_is_graphical()` now performs graphicality checks for degree sequences of simple directed graphs in linear time, an improvement from the previously used quadratic algorithm (contributed by Arnar Bjarni Arnarson @Tagl in #2537).
 - Documentation improvements.

## [0.10.10] - 2024-02-13

### Fixed

 - When `igraph_is_forest()` determined that a graph is not a directed forest, and the `roots` output parameter was set to `NULL`, it would incorrectly cache that the graph is also not an undirected forest.
 - `igraph_spanner()` now correctly ignores edge directions, and no longer crashes on directed graphs.

### Deprecated

 - `igraph_are_connected()` is renamed to `igraph_are_adjacent()`; the old name is kept available until at least igraph 1.0.

### Other

 - Documentation improvements.

## [0.10.9] - 2024-02-02

### Added

 - `igraph_is_biconnected()` checks if a graph is biconnected.
 - `igraph_realize_bipartite_degree_sequence()` constructs a bipartite graph that has the given bidegree sequence, optionally ensuring that it is connected (PR #2425 by Lára Margrét Hólmfríðardóttir @larah19).

### Fixed

 - More robust error handling in HRG code.
 - Fixed infinite loop in `igraph_hrg_sample_many()`.
 - `igraph_community_fastgreedy()` no longer crashes when providing a modularity vector only, but not a merges matrix of membership vector.
 - The graph property cache was not initialized correctly on systems where the size of `bool` was not 1 byte (#2477).
 - Compatibility with libxml2 version 2.12 (#2442).

### Deprecated

 - The macro `STR()` is deprecated; use the function `igraph_strvector_get()` instead.

### Other

 - Performance: Reduced memory usage and improved initialization performance for `igraph_strvector_t`.
 - Performance: Improved cache use by `igraph_is_bipartite()`.
 - The documentation is now also generated in Texinfo format.
 - Documentation improvements.

## [0.10.8] - 2023-11-17

### Added

 - `igraph_joint_degree_matrix()` computes the joint degree matrix, i.e. counts connections between vertices of different degrees (PR #2407 by Lára Margrét Hólmfríðardóttir @larah19).
 - `igraph_joint_degree_distribution()` computes the joint distribution of degrees at either end of edges.
 - `igraph_joint_type_distribution()` computes the joint distribution of vertex categories at either end of edges, i.e. the mixing matrix.
 - `igraph_degree_correlation_vector()` computes the degree correlation function and its various directed generalizations.

### Changed

 - The behaviour of the Pajek format reader and writer is now more closely aligned with the Pajek software and the reader is more tolerant of input it cannot interpret. Only those vertex and edge parameters are treated as valid which Pajek itself understands, therefore support for `size` is now dropped, and support for the `font` edge parameter is added. See http://mrvar.fdv.uni-lj.si/pajek/DrawEPS.htm for more information. Invalid/unrecognized parameters are now converted to igraph attributes by the reader, but just as before, they are not output by the writer.
 - The Pajek format writer now encodes newline and quotation mark characters in a Pajek-compatible manner (`\n` and `&#34;`, respectively).
 - `igraph_avg_nearest_neighbor_degree()` now supports non-simple graphs.

### Fixed

 - Resolved "ignoring duplicate libraries" warning when building tests with Xcode 15 on macOS.
 - Fixed the handling of duplicate vertex IDs in `igraph_induced_subgraph()`.
 - `igraph_vector_which_min()` and `igraph_vector_which_max()` no longer allow zero-length input, which makes them consistent with other similar functions, and was the originally intended behaviour. Passing zero-length input is invalid use and currently triggers an assertion failure.
 - `igraph_erdos_renyi_game_gnm()` and `igraph_erdos_renyi_game_gnp()` are now interruptible.
 - `igraph_de_bruijn()` and `igraph_kautz()` are now interruptible.
 - `igraph_full()`, `igraph_full_citation()`, `igraph_full_multipartite()` and `igraph_turan()` are now interruptible.
 - `igraph_avg_nearest_neighbor_degree()` did not compute `knnk` correctly in the weighted case.
 - Fixed variadic arguments of invalid types, which could cause incorrect behaviour with `igraph_matrix_print()`, as well as test suite failures, on some platforms. 32-bit x86 was affected when setting `IGRAPH_INTEGER_SIZE` to 64.
 - `igraph_subisomorphic_lad()` now returns a single null map when the pattern is the null graph.
 - `igraph_community_spinglass()` now checks its parameters more carefully.
 - `igraph_similarity_dice_pairs()` and `igraph_similarity_jaccard_pairs()` now validate vertex IDs.
 - `igraph_maxflow()` now returns an error code if the source and target vertices are the same. It used to get stuck in an infinite loop in earlier versions when the `flow` argument was non-NULL.

### Other

 - Updated vendored mini-gmp to 6.3.0.
 - `igraph_connected_components()` makes better use of the cache, improving overall performance.
 - Documentation improvements.

## [0.10.7] - 2023-09-04

### Added

 - `igraph_radius_dijkstra()` computes the graph radius with weighted edges (experimental function).
 - `igraph_graph_center_dijkstra()` computes the graph center, i.e. the set of minimum eccentricity vertices, with weighted edges (experimental function).

### Fixed

 - `igraph_full_bipartite()` now checks for overflow.
 - `igraph_bipartite_game_gnm()` and `igraph_bipartite_game_gnp()` are now more robust to overflow.
 - Bipartite graph creation functions now check input arguments.
 - `igraph_write_graph_dot()` now quotes real numbers written in exponential notation as necessary.
 - Independent vertex set finding functions could trigger the fatal error "Finally stack too large" when called on large graphs.

### Deprecated

 - `igraph_bipartite_game()` is now deprecated; use `igraph_bipartite_game_gnm()` and `igraph_bipartite_game_gnp()` instead.

### Other

 - Documentation improvements.

## [0.10.6] - 2023-07-13

### Fixed

 - Compatibility with libxml2 2.11.
 - Fixed some converge failures in `igraph_community_voronoi()`.
 - `IGRAPH_CALLOC()` and `IGRAPH_REALLOC()` now check for overflow.
 - CMake packages created with the `install` target of the CMake build system are now relocatable, i.e. the generated `igraph-targets.cmake` file does not contain absolute paths any more.

## [0.10.5] - 2023-06-29

### Added

 - `igraph_graph_power()` computes the kth power of a graph (experimental function).
 - `igraph_community_voronoi()` for detecting communities using Voronoi partitioning (experimental function).

### Changed

 - `igraph_community_walktrap()` no longer requires `modularity` and `merges` to be non-NULL when `membership` is non-NULL.
 - `igraph_isomorphic()` now supports multigraphs.
 - Shortest path related functions now consistently ignore edges with positive infinite weights.

### Fixed

 - `igraph_hub_and_authority_scores()`, `igraph_hub_score()` and `igraph_authority_score()` considered self-loops only once on the diagonal of the adjacency matrix of undirected graphs, thus the result was not identical to that obtained by `igraph_eigenvector_centrality()` on loopy undirected graphs. This is now corrected.
 - `igraph_community_infomap()` now checks edge and vertex weights for validity.
 - `igraph_minimum_spanning_tree()` and `igraph_minimum_spanning_tree_prim()` now check that edge weights are not NaN.
 - Fixed an initialization error in the string attribute combiner of the C attribute handler.
 - Fixed an issue with the weighted clique number calculation when all the weights were the same.
 - HRG functions now require a graph with at least 3 vertices; previous versions crashed with smaller graphs.
 - `igraph_arpack_rssolve()` and `igraph_arpack_rnsolve()`, i.e. the ARPACK interface in igraph, are now interruptible. As a result, several other functions that rely on ARPACK (eigenvector centrality, hub and authority scores, etc.) also became interruptible.
 - `igraph_get_shortest_paths_dijkstra()`, `igraph_get_all_shortest_paths_dijkstra()` and `igraph_get_shortest_paths_bellman_ford()` now validate the `from` vertex.
 - Fixed bugs in `igraph_local_scan_1_ecount()` for weighted undirected graphs which would miscount loops and multi-edges.

### Deprecated

- `igraph_automorphisms()` is now deprecated; its new name is `igraph_count_automorphisms()`. The old name is kept available until at least igraph 0.11.
- `igraph_hub_score()` and `igraph_authority_score()` are now deprecated. Use `igraph_hub_and_authority_scores()` instead.
- `igraph_get_incidence()` is now deprecated; its new name is `igraph_get_biadjacency()` to reflect that the returned matrix is an _adjacency_ matrix between pairs of vertices and not an _incidence_ matrix between vertices and edges. The new name is kept available until at least igraph 0.11. We plan to re-use the name in later versions to provide a proper incidence matrix where the rows are vertices and the columns are edges.
- `igraph_hrg_dendrogram()` is deprecated because it requires an attribute handler and it goes against the convention of returning attributes in vectors where possible. Use `igraph_from_hrg_dendrogram()` instead, which constructs the dendrogram as an igraph graph _and_ returns the associated probabilities in a vector.

### Other

 - Improved performance for `igraph_vertex_connectivity()`.
 - `igraph_simplify()` makes use of the cache, and avoids simplification when the graph is already known to be simple.
 - Documentation improvements.

## [0.10.4] - 2023-01-26

### Added

 - `igraph_get_shortest_path_astar()` finds a shortest path with the A* algorithm.
 - `igraph_vertex_coloring_greedy()` now supports the DSatur heuristics through `IGRAPH_COLORING_GREEDY_DSATUR` (#2284, thanks to @professorcode1).

### Changed

 - The `test` build target now only _runs_ the unit tests, but it does not _build_ them. In order to both build and run tests, use the `check` target, which continues to behave as before (PR #2291).
 - The experimental function `igraph_distances_floyd_warshall()` now has `from` and `to` parameters for choosing source and target vertices.
 - The experimental function `igraph_distances_floyd_warshall()` now has an additional `method` parameter to select a specific algorithm. A faster "Tree" variant of the Floyd-Warshall algorithm is now available (#2267, thanks to @rfulekjames).

### Fixed

 - The Bellman-Ford shortest path finder is now interruptible.
 - The Floyd-Warshall shortest path finder is now interruptible.
 - Running CTest no longer builds the tests automatically, as this interfered with VSCode, which would invoke the `ctest` executable after configuring a project in order to determine test executables. Use the `build_tests` target to build the tests first, or use the `check` target to both _build_ and _run_ all unit tests (PR #2291).

### Other

 - Improved the performance and memory usage of `igraph_widest_path_widths_floyd_warshall()`.
 - Documentation improvements.

## [0.10.3] - 2022-12-30

### Added

 - `igraph_matrix_init_array()` to initialize an igraph matrix by copying an existing C array in column-major or row-major order.
 - `igraph_layout_umap_compute_weights()` computes weights for the UMAP layout algorithm from distances. This used to be part of `igraph_layout_umap()`, but it is now in a separate function to allow the user to experiment with different weighting schemes.
 - `igraph_triangular_lattice()` to generate triangular lattices of various kinds (#2235, thanks to @rfulekjames).
 - `igraph_hexagonal_lattice()` to generate hexagonal lattices of various kinds (#2262, thanks to @rfulekjames).
 - `igraph_tree_from_parent_vector()` to create a tree or a forest from a parent vector (i.e. a vector that encodes the parent vertex of each vertex).
 - `igraph_induced_subgraph_edges()` produces the IDs of edges contained within a subgraph induced by the given vertices.

### Changed

 - The signature of the experimental `igraph_layout_umap()` function changed; the last argument is now a Boolean that specifies whether distances should already be treated as weights, and the sampling probability argument was removed.

### Fixed

 - `igraph_transitivity_barrat()`, `igraph_community_fluid_communities()`, `igraph_sir()`, `igraph_trussness()` and graphlet functions did not correctly detect when a directed input graph had effective multi-edges due to ignoring edge directions. Such graphs are now rejected by these functions.
 - Fixed a bug in `igraph_2dgrid_move()` that sometimes crashed the Large Graph Layout function when a grid cell became empty.
 - `igraph_pagerank()` and `igraph_personalized_pagerank()` would fail to converge when the ARPACK implementation was used and a vertex had more than one outgoing edge but all these edges had zero weights.
 - `igraph_pagerank()` and `igraph_personalized_pagerank()` no longer allow negative weights. Previously, edges with negative weights were silently ignored when using the PRPACK implementation. The ARPACK implementation would issue a warning saying that they are ignored, but in fact it computed an incorrect result.
 - `igraph_all_st_cuts()` and `igraph_all_st_mincuts()` no longer trigger the "Finally stack too large" fatal error when called on certain large graphs. This was a regression in igraph 0.10.
 - `igraph_community_label_propagation()` no longer rounds weights to integers. This was a regression in igraph 0.10.
 - `igraph_read_graph_graphdb()` does more thorough checks on the input file.
 - `igraph_calloc()` did not zero-initialize the allocated memory. This is now corrected. Note that the macro `IGRAPH_CALLOC()` was _not_ affected.
 - Fixed new warnings issued by the Xcode 14.1 toolchain.

### Deprecated

- `igraph_subgraph_edges()` is now deprecated to avoid confusion with `igraph_induced_subgraph_edges()`; its new name is `igraph_subgraph_from_edges()`. The old name is kept available until at least igraph 0.11.

### Other

 - Significantly improved performance for `igraph_matrix_transpose()`.
 - Documentation improvements.

## [0.10.2] - 2022-10-14

### Added

 - `igraph_distances_cutoff()` and `igraph_distances_dijkstra_cutoff()` calculate shortest paths with an upper limit on the path length (experimental functions).
 - `igraph_distances_floyd_warshall()` for computing all-pairs shortest path lengths in dense graphs (experimental function).
 - `igraph_ecc()` computes the edge clustering coefficient of some edges (experimental function).
 - `igraph_voronoi()` computes a Voronoi partitioning of vertices (experimental function).
 - `igraph_count_multiple_1()` determines the multiplicity of a single edge in the graph.
 - `igraph_dqueue_get()` accesses an element in a queue by index.
 - `igraph_degree_1()` efficiently retrieves the degee of a single vertex.
 - `igraph_lazy_adjlist_has()` and `igraph_lazy_inclist_has()` to check if adjacent vertices / incident edges have already been computed and stored for a given vertex in a lazy adjlist / inclist.

### Changed

 - `igraph_edge()` now verifies that the input edge ID is valid.
 - `igraph_community_leading_eigenvector()`, `igraph_adjacency_spectral_embedding()`, `igraph_laplacian_spectral_embedding()`, `igraph_arpack_rssolve()` and `igraph_arpack_rnsolve()` now generate a random starting vector using igraph's own RNG if needed instead of relying on LAPACK or ARPACK to do so. This makes sure that the results obtained from these functions remain the same if igraph's RNG is seeded with the same value.
 - `igraph_community_leading_eigenvector()` does not stop the splitting process any more when there are multiple equally likely splits (indicated by the multiplicity of the leading eigenvector being larger than 1). The algorithm picks an arbitrary split instead and proceeds normally.

### Fixed

 - Fixed a bug in `igraph_get_k_shortest_paths()` that sometimes yielded incorrect results on undirected graphs when the `mode` argument was set to `IGRAPH_OUT` or `IGRAPH_IN`.
 - `igraph_trussness()` is now interruptible.
 - `igraph_spanner()` is now interruptible.
 - `igraph_layout_umap()` and `igraph_layout_umap3d()` are now interruptible.
 - In some rare cases, roundoff errors would cause `igraph_distance_johnson()` to fail on graphs with negative weights.
 - `igraph_eulerian_cycle()` and `igraph_eulerian_path()` now returns a more specific error code (`IGRAPH_ENOSOL`) when the graph contains no Eulerian cycle or path.
 - `igraph_heap_init_array()` did not copy the array data correctly for non-real specializations.
 - `igraph_layout_umap_3d()` now actually uses three dimensions.
 - `igraph_layout_umap()` and `igraph_layout_umap_3d()` are now interruptible.
 - `igraph_vit_create()` and `igraph_eit_create()` no longer fails when trying to create an iterator for the null graph or edgeless graph from an empty range-based vertex or edge selector.
 - `igraph_write_graph_leda()` did not correctly print attribute names in some warning messages.
 - Addressed new warnings introduced by Clang 15.
 - In the generated pkg-config file, libxml2 is now placed in the `Requires.private` section instead of the `Libs.private` one.

### Removed

 - Removed unused and undocumented `igraph_bfgs()` function.
 - Removed the undocumented function `igraph_complex_mod()`. Use `igraph_complex_abs()` instead, as it has identical functionality.

### Deprecated

 - The `IGRAPH_EDRL` error code was deprecated; the DrL algorithm now returns `IGRAPH_FAILURE` when it used to return `IGRAPH_EDRL` (not likely to happen in practice).
 - The undocumented function `igraph_dqueue_e()` is now deprecated and replaced by `igraph_dqueue_get()`.
 - `igraph_finite()`, `igraph_is_nan()`, `igraph_is_inf()`, `igraph_is_posinf()` and `igraph_is_neginf()` are now deprecated. They were relics from a time when no standard alternatives existed. Use the C99 standard `isfinite()`, `isnan()` and `isinf()` instead.

### Other

 - Documentation improvements.

## [0.10.1] - 2022-09-08

### Fixed

 - Corrected a regression (compared to igraph 0.9) in weighted clique search functions.
 - `igraph_girth()` no longer fails when the graph has no cycles and the `girth` parameter is set to `NULL`.
 - `igraph_write_graph_gml()` did not respect entity encoding options when writing the `Creator` line.
 - Fixed potential memory leak on out-of-memory condition in `igraph_asymmetric_preference_game()`, `igraph_vs_copy()` and `igraph_es_copy()`.
 - Fixed an assertion failure in `igraph_barabasi_game()` and `igraph_barabasi_aging_game()` when passing in negative degree exponents.
 - Fixed a compilation failure with some old Clang versions.

### Changed

 - `igraph_write_graph_leda()` can now write boolean attributes.

### Other

 - Support for ARM64 on Windows.
 - Documentation improvements.

## [0.10.0] - 2022-09-05

### Release notes

This release focuses on infrastructural improvements, stability, and making the igraph interface more consistent, more predictable and easier to use. It contains many API-breaking changes and function renamings, in preparation for a future 1.0 release, at which point the API will become stable. Changes in this direction are likely to continue through a 0.11 release. It is recommended that you migrate your code from 0.9 to 0.10 soon, to make the eventual transition to 1.0 easier.

Some of the highlights are:

 - A consistent use of `igraph_integer_t` for all indices and most integer quantities, both in the API and internally. This type is 64-bit by default on all 64-bit systems, bringing support for very large graphs with more than 2 billion vertices. Previously, vertex and edge indices were often represented as `igraph_real_t`. The move to an `igraph_integer_t` also implies a change from `igraph_vector_t` to `igraph_vector_int_t` in many functions.
 - The random number generation framework has been overhauled. Sampling from the full range of `igraph_integer_t` is now possible. Similarly, the sampling of random reals has been improved to utilize almost the full range of the mantissa of an `igraph_real_t`.
 - There is a new fully memory-managed container type for lists of vectors (`igraph_vector_list_t`), replacing most previous uses of the non-managed `igraph_vector_ptr_t`. Functions that previously used `igraph_vector_ptr_t` to return results and relied on the user to manage memory appropriately are now using `igraph_vector_list_t`, `igraph_graph_list_t` or similar and manage memory on their own.
 - Some simple graph properties, such as whether a graph contains self-loops or multi-edges, or whether it is connected, are now cached in the graph data structure. Querying these properties for a second time will take constant computational time. The `igraph_invalidate_cache()` function is provided for debugging purposes. It will invaidate all cache entries.
 - File format readers are much more robust and more tolerant of invalid input.
 - igraph is much more resilient to overflow errors.
 - Many improvements to robustness and reliability, made possible by internal refactorings.

### Breaking changes

 - igraph now requires CMake 3.18 or later.
 - In order to facilitate the usage of graphs with more than 2 billion vertices and edges, we have made the size of the `igraph_integer_t` data type to be 32 bits on 32-bit platforms and 64 bits on 64-bit platforms by default. You also have the option to compile a 32-bit igraph variant on a 64-bit platform by changing the `IGRAPH_INTEGER_SIZE` build variable in CMake to 32.
 - `igraph_bool_t` is now a C99 `bool` and not an `int`. Similarly, `igraph_vector_bool_t` now consumes `sizeof(bool)` bytes per entry only, not `sizeof(int)`. The standard constants `true` and `false` may be used for Boolean values for readability.
 - The random number generator interface, `igraph_rng_type_t`, has been overhauled. Check the declaration of the type for details.
 - The default random number generator has been changed from Mersenne Twister to PCG32.
 - Functions related to spectral coarse graining (i.e. all functions starting with `igraph_scg_...`) were separated into a project of its own. If you wish to keep on using these functions, please refer to the repository hosting the spectral coarse graining code at https://github.com/igraph/igraph-scg . The spectral coarse graining code was updated to support igraph 0.10.
 - Since `igraph_integer_t` aims to be the largest integer size that is feasible on a particular platform, there is no need for generic data types based on `long int` any more. The `long` variants of generic data types (e.g., `igraph_vector_long_t`) are therefore removed; you should use the corresponding `int` variant instead, whose elements are of type `igraph_integer_t`.
 - Generic data types based on `float` were removed as they were not used anywhere in the library.
 - Several igraph functions that used to take a `long int` or return a `long int` now takes or returns an `igraph_integer_t` instead to make the APIs more consistent. Similarly, igraph functions that used `igraph_vector_t` for arguments that take or return _integral_ vectors (e.g., vertex or edge indices) now take `igraph_vector_int_t` instead. Graph-related functions where the API was changed due to this reason are listed below, one by one.
 - Similarly, igraph functions that used to accept the `long` variant of a generic igraph data type (e.g., `igraph_vector_long_t`) now take the `int` variant of the same data type.
 - The type `igraph_stack_ptr_t` and its associated functions were removed. Use `igraph_vector_ptr_t` and associated functions instead.
 - Error handlers should no longer perform a `longjmp()`. Doing so will introduce memory leaks, as resource cleanup is now done in multiple stages, through multiple calls to the error handler. Thus, the error handler should either abort execution immediately (as the default handler does), or report the error, call `IGRAPH_FINALLY_FREE()`, and return normally.
 - Most callback functions now return an error code. In previous versions they returned a boolean value indicating whether to terminate the search. A request to stop the search is now indicated with the special return code `IGRAPH_STOP`.
 - `igraph_add_edges()` now uses an `igraph_vector_int_t` for its `edges` parameter.
 - `igraph_adjacency()` no longer accepts a negative number of edges in its adjacency matrix. When negative entries are found, an error is generated.
 - `igraph_adjacency()` gained an additional `loops` argument that lets you specify whether the diagonal entries should be ignored or should be interpreted as raw edge counts or _twice_ the number of edges (which is common in linear algebra contexts).
 - `igraph_all_minimal_st_separators()` now returns the separators in an `igraph_vector_int_list_t` containing `igraph_vector_int_t` vectors.
 - `igraph_all_st_cuts()` and `igraph_all_st_mincuts()` now return the cuts in an `igraph_vector_int_list_t` containing `igraph_vector_int_t` vectors.
 - `igraph_arpack_unpack_complex()` now uses `igraph_integer_t` for its `nev` argument instead of `long int`.
 - `igraph_articulation_points()` now uses an `igraph_vector_int_t` to return the list of articulation points, not an `igraph_vector_t`.
 - `igraph_assortativity_nominal()` now accepts vertex types in an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - `igraph_asymmetric_preferennce_game()` now uses an `igraph_vector_int_t` to return the types of the nodes in the generated graph.
 - `igraph_atlas()` now uses `igraph_integer_t` for its `number` argument.
 - `igraph_automorphism_group()` now returns the generators in an `igraph_vector_int_list_t` instead of a pointer vector containing `igraph_vector_t` objects.
 - `igraph_barabasi_game()`, `igraph_barabasi_aging_game()`, `igraph_recent_degree_game()` and `igraph_recent_degree_aging_game()` now use an `igraph_vector_int_t` for the out-degree sequence of the nodes being generated instead of an `igraph_vector_t`.
 - `igraph_bfs()` now takes an `igraph_vector_int_t` for its `roots`, `restricted`, `order`, `father`, `pred`, `succ` and `dist` arguments instead of an `igraph_vector_t`.
 - `igraph_bfs_simple()` now takes `igraph_vector_int_t` for its `vids`, `layers` and `parents` arguments instead of an `igraph_vector_t`.
 - `igraph_bfs_simple()` now returns -1 in `parents` for the root node of the traversal, and -2 for unreachable vertices. This is now consistent with other functions that return a parent vector.
 - `igraph_biconnected_components()` now uses an `igraph_vector_int_t` to return the list of articulation points, not an `igraph_vector_t`. Also, the container used for the edges and vertices of the components is now an `igraph_vector_int_list_t` instead of a pointer vector containing `igraph_vector_t` objects.
 - `igraph_bipartite_projection()` now uses `igraph_vector_int_t` to return `multiplicity1` and `multiplicity2`, not `igraph_vector_t`.
 - `igraph_bridges()` now uses an `igraph_vector_int_t` to return the list of bridges, not an `igraph_vector_t`.
 - `igraph_callaway_traits_game()` returns the node types in an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - `igraph_canonical_permutation()` now uses an `igraph_vector_int_t` for its labeling parameter.
 - `igraph_cattribute_list()` now uses `igraph_vector_int_t` to return `gtypes`, `vtypes` and `etypes`.
 - `igraph_cited_type_game()` now uses an `igraph_vector_int_t` for its types parameter.
 - `igraph_citing_cited_type_game()` now uses an `igraph_vector_int_t` for its
   types parameter.
 - `igraph_clique_handler_t` now uses an `igraph_vector_int_t` for its `clique` parameter, and must return an `igraph_error_t`. Use `IGRAPH_STOP` as the return code to terminate the search prematurely. The vector that the handler receives is owned by the clique search routine. If you want to hold on to the vector for a longer period of time, you need to make a copy of it in the handler. Cliques passed to the callback are marked as `const` as a reminder to this change.
 - The `res` parameter of `igraph_cliques()` is now an `igraph_vector_int_list_t`.
 - Callbacks used by `igraph_cliques_callback()` need to be updated to account for the fact that the callback does not own the clique passed to it any more; the callback needs to make a copy if it wants to hold on to the clique for a longer period of time. If the callback does not need to store the clique, it does not need to do anything any more, and it must not destroy or free the clique.
 - `igraph_closeness()` and `igraph_closeness_cutoff()` now use an `igraph_vector_int_t` to return `reachable_count`, not an `igraph_vector_t`.
 - `igraph_cohesive_blocks()` now uses an `igraph_vector_int_t` to return the mapping from block indices to parent block indices, and the `cohesion`; also, it uses an `igraph_vector_int_list_t` to return the blocks themselves instead of a pointer vector of `igraph_vector_t`.
 - The `igraph_community_eb_get_merges()` bridges parameter now starts the indices into the edge removal vector at 0, not 1.
 - The `igraph_community_eb_get_merges()` now reports an error when not all edges in the graph are removed, instead of a nonsensical result.
 - `igraph_community_edge_betweenness()` now uses an `igraph_vector_int_t` to return the edge IDs in the order of their removal as well as the list of edge IDs whose removal broke a single component into two.
 - `igraph_community_fluid_communities()` does not provide the modularity in a separate output argument any more; use `igraph_modularity()` to retrieve the modularity if you need it.
 - `igraph_community_infomap()` now uses `igraph_integer_t` for its `nb_trials` argument.
 - `igraph_community_label_propagation()` now uses an `igraph_vector_int_t` for its `initial` parameter. It also takes a `mode` argument that specifies how labels should be propagated along edges (forward, backward or ignoring edge directions).
 - `igraph_community_label_propagation()` does not provide the modularity in a separate output argument any more; use `igraph_modularity()` to retrieve the modularity if you need it.
 - `igraph_community_leiden()` has an additional parameter to indicate the number of iterations to perform (PR #2177).
 - `igraph_community_walktrap()`, `igraph_community_edge_betweenness()`, `igraph_community_eb_get_merges()`, `igraph_community_fastgreedy()`, `igraph_community_to_membership()`, `igraph_le_community_to_membership()`, `igraph_community_leading_eigenvector()` now use an `igraph_vector_int_t` for their `merges` parameter.
 - `igraph_community_walktrap()` now uses `igraph_integer_t` for its `steps` argument.
 - `igraph_coreness()` now uses an `igraph_vector_int_t` to return the coreness
   values.
 - `igraph_convex_hull()` now uses an `igraph_vector_int_t` to return the indices of the input vertices that were chosen to be in the convex hull.
 - `igraph_correlated_game()` and `igraph_correlated_pair_game()` now take an `igraph_vector_int_t` as the permutation vector, not an `igraph_vector_t`.
 - `igraph_create()` now uses an `igraph_vector_int_t` for its `edges` parameter.
 - `igraph_create_bipartite()` now uses an `igraph_vector_int_t` for its `edges` parameter.
 - `igraph_compose()` now returns the edge maps in an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - `igraph_count_multiple()` now returns the multiplicities in an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - `igraph_decompose()` now uses an `igraph_integer_t` for its `maxcompno` and `minelements` arguments instead of a `long int`.
 - `igraph_degree()` now uses an `igraph_vector_int_t` to return the degrees.  If you need the degrees in a vector containing floating-point numbers instead (e.g., because you want to pass them on to some other function that takes an `igraph_vector_t`), use `igraph_strength()` instead with a null weight vector.
 - `igraph_degree_sequence_game()` now takes degree sequences represented as `igraph_vector_int_t` instead of `igraph_vector_t`.
 - `igraph_degseq_t`, used by `igraph_degree_sequence_game()`, uses new names for its constants. The old names are deprecated, but retained for compatibility.  See `igraph_constants.h` to see which new name corresponds to which old one.
 - `igraph_delete_vertices_idx()` now uses `igraph_vector_int_t` vectors to return the mapping and the inverse mapping of old vertex IDs to new ones.
 - `igraph_deterministic_optimal_imitation()` now expects the list of strategies in an `igraph_vector_int_t` instead of an `igraph_int_t`.
 - `igraph_dfs()` now takes an `igraph_vector_int_t` for its `order`, `order_out`, `father` and `dist` arguments instead of an `igraph_vector_t`. Furthermore, these vectors will contain -2 for vertices that have not been visited; in earlier versions, they used to contain NaN instead. Note that -1 is still used in the `father` vector to indicate the root of a DFS tree.
 - `igraph_diameter()` and `igraph_diameter_dijkstra()` now use `igraph_vector_int_t` vectors to return the list of vertex and edge IDs in the diameter.
 - `igraph_dominator_tree()` now takes an `igraph_vector_int_t` for its `dom` and `leftout` arguments instead of an `igraph_vector_t`.
 - `igraph_dyad_census()` now uses `igraph_real_t` instead of `igraph_integer_t` for its output arguments, and it no longer returns -1 when overflow occurs.
 - `igraph_edges()` now takes an `igraph_vector_int_t` for its `edges` argument instead of an `igraph_vector_t`.
 - `igraph_es_multipairs()` was removed; you can use the newly added `igraph_es_all_between()` instead.
 - `igraph_establishment_game()` now takes an `igraph_vector_int_t` for its `node_type_vec` argument instead of an `igraph_vector_t`.
 - `igraph_eulerian_path()` and `igraph_eulerian_cycle()` now use `igraph_vector_int_t` to return the list of edge and vertex IDs participating in an Eulerian path or cycle instead of an `igraph_vector_t`.
 - `igraph_feedback_arc_set()` now uses an `igraph_vector_int_t` to return the IDs of the edges in the feedback arc set instead of an `igraph_vector_t`.
 - `igraph_get_adjacency()` no longer has the `eids` argument, which would produce an adjacency matrix where non-zero values were 1-based (not 0-based) edge IDs. If you need a matrix with edge IDs, create it manually.
 - `igraph_get_adjacency_sparse()` now returns the sparse adjacency matrix in an `igraph_sparsemat_t` structure, and it assumes that the input matrix is _initialized_ for sake of consistency with other igraph functions.
 - `igraph_get_adjacency()` and `igraph_get_adjacency_sparse()` now has a `loops` argument that lets the user specify how loop edges should be handled.
 - `igraph_get_edgelist()` now uses an `igraph_vector_int_t` for its `res` parameter.
 - `igraph_get_eids()` now uses `igraph_vector_int_t` to return lists of edge IDs and to receive lists of vertex IDs.
 - The `path` argument of `igraph_get_eids()` was removed. You can replicate the old behaviour by constructing the list of vertex IDs explicitly from the path by duplicating each vertex in the path except the first and last ones. A helper function called `igraph_expand_path_to_pairs()` is provided to ease the transition.
 - `igraph_get_eids_multi()` was removed as its design was fundamentally broken; there was no way to retrieve the IDs of all edges between a specific pair of vertices without knowing in advance how many such edges there are in the graph.  Use `igraph_get_all_eids_between()` instead.
 - `igraph_get_incidence()` now returns the vertex IDs corresponding to the rows and columns of the incidence matrix as `igraph_vector_int_t`.
 - `igraph_get_shortest_path()`, `igraph_get_shortest_path_bellman_ford()` and `igraph_get_shortest_path_dijkstra()` now use `igraph_vector_int_t` vectors to return the list of vertex and edge IDs in the shortest path.
 - `igraph_get_shortest_paths()`, `igraph_get_shortest_paths_dijkstra()` and `igraph_get_shortest_paths_bellman_ford()` now use an `igraph_vector_int_t` to return the predecessors and inbound edges instead of an `igraph_vector_long_t`.
 - The functions `igraph_get_all_shortest_paths()`, `igraph_get_all_shortest_paths_dijkstra()`, `igraph_get_shortest_paths()`, `igraph_get_shortest_paths_bellman_ford()` and `igraph_get_shortest_paths_dijkstra()` now return paths in an `igraph_vector_int_list_t` instead of a pointer vector containing `igraph_vector_t` objects.
 - The vector of parents in `igraph_get_shortest_paths()`, `igraph_get_shortest_paths_bellman_ford()` and `igraph_get_shortest_paths_dijkstra()` now use -1 to represent the starting vertex, and -2 for unreachable vertices.
 - The `maps` parameters in `igraph_get_isomorphisms_vf2()` and `igraph_get_subisomorphisms_vf2()` are now of type `igraph_vector_int_list_t`.
 - `igraph_get_stochastic()` now has an additional `weights` argument for edge weights.
 - `igraph_get_stochastic_sparse()` now returns the sparse adjacency matrix in an `igraph_sparsemat_t` structure, and it assumes that the input matrix is _initialized_ for sake of consistency with other igraph functions. It also received an additional `weights` argument for edge weights.
 - `igraph_girth()` now uses an `igraph_vector_int_t` for its `circle` parameter.
 - `igraph_girth()` now uses `igraph_real_t` as the return value so we can return infinity for graphs with no cycles (instead of zero).
 - The `cliques` parameters of type `igraph_vector_ptr_t` in `igraph_graphlets()`, `igraph_graphlets_candidate_basis()` and `igraph_graphlets_project()` were changed to an `igraph_vector_int_list_t`.
 - `igraph_hrg_init()` and `igraph_hrg_resize()` now takes an `igraph_integer_t` as their size arguments instead of an `int`.
 - `igraph_hrg_consensus()` now returns the parent vector in an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - `igraph_hrg_create()` now takes a vector of probabilities corresponding to the internal nodes of the dendogram. It used to also take probabilities for the leaf nodes and then ignore them.
 - `igraph_hrg_predict()` now uses an `igraph_vector_int_t` for its `edges` parameter.
 - `igraph_hrg_sample()` now always samples a single graph only. Use `igraph_hrg_sample_many()` if you need more than one sample, and call `igraph_hrg_fit()` beforehand if you do not have a HRG model but only a single input graph.
 - `igraph_hrg_size()` now returns an `igraph_integer_t` instead of an `int`.
 - `igraph_incidence()` does not accept negative incidence counts any more.
 - `igraph_incident()` now uses an `igraph_vector_int_t` for its `eids` parameter.
 - The `res` parameter in `igraph_independent_vertex_sets()` is now an `igraph_vector_int_list_t`.
 - `igraph_induced_subgraph_map()` now uses `igraph_vector_int_t` vectors to return the mapping and the inverse mapping of old vertex IDs to new ones.
 - `igraph_intersection()` now uses an `igraph_vector_int_t` for its `edge_map1` and `edge_map2` parameters.
 - The `edgemaps` parameter of `igraph_intersection_many()` is now an `igraph_vector_int_list_t` instead of a pointer vector.
 - `igraph_is_chordal()` now uses an `igraph_vector_int_t` for its `alpha`, `alpham1` and `fill_in` parameters.
 - `igraph_is_graphical()` and `igraph_is_bigraphical()` now take degree sequences represented as `igraph_vector_int_t` instead of `igraph_vector_t`.
 - `igraph_is_matching()`, `igraph_is_maximal_matching()` and `igraph_maximum_bipartite_matching` now use an `igraph_vector_int_t` to return the matching instead of an `igraph_vector_long_t`.
 - `igraph_is_mutual()` has an additional parameter which controls whether directed self-loops are considered mutual.
 - The `vids` parameter for `igraph_isoclass_subgraph()` is now an `igraph_vector_int_t` instead of `igraph_vector_t`.
 - `igraph_isomorphic_vf2()`, `igraph_get_isomorphisms_vf2_callback()` (which used to be called `igraph_isomorphic_function_vf2()`) and `igraph_isohandler_t` now all use `igraph_vector_int_t` for their `map12` and `map21` parameters.
 - The `cliques` parameter of type `igraph_vector_ptr_t` in `igraph_largest_cliques()` was changed to an `igraph_vector_int_list_t`.
 - The `res` parameters of type `igraph_vector_ptr_t` in `igraph_largest_independent_vertex_sets()` and `igraph_largest_weighted_cliques()` were changed to an `igraph_vector_int_list_t`.
 - The dimension vector parameter for `igraph_square_lattice()` (used to be `igraph_lattice()`) is now an `igraph_vector_int_t` instead of `igraph_vector_t`.
 - The maxiter parameter of `igraph_layout_bipartite()` is now an `igraph_integer_t` instead of `long int`.
 - The fixed parameter of `igraph_layout_drl()` and `igraph_layout_drl_3d()` was removed as it has never been implemented properly.
 - The width parameter of `igraph_layout_grid()` is now an `igraph_integer_t` instead of `long int`.
 - The width and height parameters of `igraph_layout_grid_3d()` are now `igraph_integer_t` instead of `long int`.
 - The dimension parameter of `igraph_layout_mds()` is now an `igraph_integer_t` instead of `long int`.
 - The `roots` and `rootlevel` parameters of `igraph_layout_reingold_tilford()` are now `igraph_vector_int_t` instead of `igraph_vector_t`.
 - The `roots` and `rootlevel` parameters of `igraph_layout_reingold_tilford_circular()` are now `igraph_vector_int_t` instead of `igraph_vector_t`.
 - The order parameter of `igraph_layout_star()` is now an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - The maxiter parameter of `igraph_layout_sugiyama()` is now an `igraph_integer_t` instead of `long int`. Also, the function now uses an `igraph_vector_int_t` for its `extd_to_orig_eids` parameter.
 - The shifts parameter of `igraph_lcf_vector()` is now an `igraph_vector_int_t` instead of an `igraph_vector_t`.
 - `igraph_matrix_minmax()`, `igraph_matrix_which_minmax()`, `igraph_matrix_which_min()` and `igraph_matrix_which_max()` no longer return an error code. The return type is now `void`. These functions never fail.
 - `igraph_maxflow()` now uses an `igraph_vector_int_t` for its `cut`, `partition` and `partition2` parameters.
 - The `igraph_maxflow_stats_t` struct now contains `igraph_integer_t` values instead of `int` ones.
 - The `res` parameters in `igraph_maximal_cliques()` and `igraph_maximal_cliques_subset()` are now of type `igraph_vector_int_list_t`.
 - Callbacks used by `igraph_maximal_cliques_callback()` need to be updated to account for the fact that the callback does not own the clique passed to it any more; the callback needs to make a copy if it wants to hold on to the clique for a longer period of time. If the callback does not need to store the clique, it does not need to do anything any more, and it must not destroy or free the clique.
 - The `res` parameter in `igraph_maximal_independent_vertex_sets()` is now an `igraph_vector_int_list_t`.
 - `igraph_maximum_cardinality_search()` now uses an `igraph_vector_int_t` for its `alpha` and `alpham1` arguments.
 - `igraph_mincut()` now uses an `igraph_vector_int_t` for its `cut`, `partition` and `partition2` parameters.
 - `igraph_moran_process()` now expects the list of strategies in an `igraph_vector_int_t` instead of an `igraph_int_t`.
 - Motif callbacks of type `igraph_motifs_handler_t` now take an `igraph_vector_int_t` with the vertex IDs instead of an `igraph_vector_t`, and use `igraph_integer_t` for the isoclass parameter.
 - Motif functions now use `igraph_integer_t` instead of `int` for their `size` parameter.
 - `igraph_neighborhood_size()` now uses an `igraph_vector_int_t` for its `res` parameter.
 - The `res` parameter of `igraph_neighborhood()` is now an `igraph_vector_int_list_t`.
 - `igraph_neighbors()` now uses an `igraph_vector_int_t` for its `neis` parameter.
 - `igraph_permute_vertices()` now takes an `igraph_vector_int_t` as the permutation vector.
 - `igraph_power_law_fit()` does not calculate the p-value automatically any more because the previous estimation method did not match the results from the original paper of Clauset, Shalizi and Newman (2009) and the implementation of the method outlined in the paper runs slower than the previous naive estimate. A separate function named `igraph_plfit_result_calculate_p_value()` is now provided for calculating the p-value. The automatic selection of the `x_min` cutoff also uses a different method than earlier versions. As a consequence, results might be slightly different if you used tests where the `x_min` cutoff was selected automatically. The new behaviour is now consistent with the defaults of the underlying `plfit` library.
 - `igraph_preference_game()` now uses an `igraph_vector_int_t` to return the types of the nodes in the generated graph.
 - `igraph_random_walk()` now uses an `igraph_vector_int_t` for its results. Also, the function now takes both vertices and edges as parameters. It can return IDs of vertices and/or edges on the walk.  The function now takes weights as a parameter to support weighted graphs.
 - `igraph_random_edge_walk()` now uses an `igraph_vector_int_t` for its `edgewalk` parameter.
 - `igraph_read_graph_dimacs_flow()` now uses an `igraph_vector_int_t` for its label parameter.
 - `igraph_read_graph_graphml()` now uses `igraph_integer_t` for its `index` argument.
 - `igraph_read_graph_pajek()` now creates a Boolean `type` attribute for bipartite graphs.  Previously it created a numeric attribute.
 - `igraph_realize_degree_sequence()` now uses an `igraph_vector_int_t` for its `outdeg` and `indeg` parameters.
 - `igraph_reindex_membership()` now uses an `igraph_vector_int_t` for its `new_to_old` parameter.
 - `igraph_rng_seed()` now requires an `igraph_uint_t` as its seed arguments. RNG implementations are free to use only the lower bits of the seed if they do not support 64-bit seeds.
 - `igraph_rngtype_rand` (i.e. the RNG that is based on BSD `rand()`) was removed due to poor statistical properties that sometimes resulted in weird artifacts like all-even "random" numbers when igraph's usage patterns happened to line up with the shortcomings of the `rand()` generator in a certain way.
 - `igraph_roulette_wheel_imitation()` now expects the list of strategies in an `igraph_vector_int_t` instead of an `igraph_int_t`.
 - `igraph_similarity_dice_pairs()` now uses an `igraph_vector_int_t` for its `pairs` parameter.
 - `igraph_similarity_jaccard_pairs()` now uses an `igraph_vector_int_t` for its `pairs` parameter.
 - `igraph_simple_interconnected_islands_game()` does not generate multi-edges between islands any more.
 - `igraph_sort_vertex_ids_by_degree()` and `igraph_topological_sorting()` now use an `igraph_vector_int_t` to return the vertex IDs instead of an `igraph_vector_t`.
 - `igraph_spanning_tree()`, `igraph_minimum_spanning_tree()` and `igraph_random_spanning_tree()` now all use an `igraph_vector_int_t` to return the vector of edge IDs in the spanning tree instead of an `igraph_vector_t`.
 - `igraph_sparsemat_cholsol()`, `igraph_sparsemat_lusol()`, `igraph_sparsemat_symbqr()` and `igraph_sparsemat_symblu()` now take an `igraph_integer_t` as their `order` parameter.
 - `igraph_sparsemat_count_nonzero()` and `igraph_sparsemat_count_nonzerotol()` now return an `igraph_integer_t`.
 - `igraph_sparsemat_is_symmetric()` now returns an error code and the result itself is provided in an output argument.
 - The `values` argument of `igraph_sparsemat_transpose()` was removed; now the function always copies the values over to the transposed matrix.
 - `igraph_spmatrix_t` and related functions were removed as they mostly duplicated functionality that was already present in `igraph_sparsemat_t`.  Functions that used `igraph_spmatrix_t` in the library now use `igraph_sparsemat_t`.
 - `igraph_stochastic_imitation()` now expects the list of strategies in an `igraph_vector_int_t` instead of an `igraph_int_t`.
 - `igraph_st_mincut()` now uses an `igraph_vector_int_t` for its `cut`, `partition` and `partition2` parameters.
 - `igraph_st_vertex_connectivity()` now ignores edges between source and target for `IGRAPH_VCONN_NEI_IGNORE`
 - `igraph_strvector_get()` now returns strings in the return value, not in an output argument.
 - `igraph_subcomponent()` now uses an `igraph_integer_t` for the seed vertex instead of an `igraph_real_t`. It also uses an `igraph_vector_int_t` to return the list of vertices in the same component as the seed vertex instead of an `igraph_vector_t`.
 - `igraph_subisomorphic_vf2()`, `igraph_get_subisomorphisms_vf2_callback()` (which used to be called `igraph_subisomorphic_function_vf2()`) and `igraph_isomorphic_bliss()` now all use `igraph_vector_int_t` for their `map12` and `map21` parameters.
 - The `maps` parameters in `igraph_subisomorphic_lad()`, `igraph_get_isomorphisms_vf2()` and `igraph_get_subisomorphisms_vf2()` are now of type `igraph_vector_int_list_t`.
 - `igraph_subisomorphic_lad()` now uses an `igraph_vector_int_t` for its `map` parameter. Also, its `domains` parameter is now an `igraph_vector_int_list_t` instead of a pointer vector containing `igraph_vector_t` objects.
 - `igraph_unfold_tree()` now uses an `igraph_vector_int_t` for its `vertex_index` and `roots` parameters.
 - `igraph_union()` now uses an `igraph_vector_int_t` for its `edge_map1` and `edge_map2` parameters.
 - The `edgemaps` parameter of `igraph_union_many()` is now an `igraph_vector_int_list_t` instead of a pointer vector.
 - `igraph_vector_init_copy()` was refactored to take _another_ vector that the newly initialized vector should copy. The old array-based initialization function is now called `igraph_vector_init_array()`.
 - `igraph_vector_ptr_init_copy()` was renamed to `igraph_vector_ptr_init_array()` for sake of consistency.
 - `igraph_vs_vector()`, `igraph_vss_vector()` and `igraph_vs_vector_copy()` now all take an `igraph_vector_int_t` as the vector of vertex IDs, not an `igraph_vector_t`. Similarly, `igraph_vs_as_vector()` now returns the vector of matched vertex IDs in an `igraph_vector_int_t`, not an `igraph_vector_t`.
 - The `res` parameter of `igraph_weighted_cliques()` is now an `igraph_vector_int_list_t`.
 - `igraph_write_graph_dimacs_flow()` now uses `igraph_integer_t` for the source and target vertex index instead of a `long int`.
 - `igraph_vector_*()`, `igraph_matrix_*()`, `igraph_stack_*()`, `igraph_array_*()` and several other generic igraph data types now use `igraph_integer_t` for indexing, _not_ `long int`. Please refer to the headers for the exact details; the list of affected functions is too large to include here.
 - `igraph_vector_minmax()` and `igraph_vector_which_minmax()` no longer return an error code. The return type is now `void`. These functions never fail.
 - `igraph_vector_order()` was removed; use `igraph_vector_int_pair_order()` instead. (The original function worked for vectors containing integers only).
 - `igraph_vector_resize_min()` and `igraph_matrix_resize_min()` no longer return an error code (return type is now `void`). The vector or matrix is always left in a consistent state by these functions, with all data intact, even if releasing unused storage is not successful.
 - `igraph_vector_qsort_ind()` and its variants now take an `igraph_order_t` enum instead of a boolean to denote whether the order should be ascending or descending.
 - `igraph_weighted_adjacency()` now returns the weights in a separate vector instead of storing it in a vertex attribute. The reason is twofold: first, the previous solution worked only with the C attribute handler (not the ones from the higher-level interfaces), and second, it wasn't consistent with other igraph functions that use weights provided as separate arguments.
 - The `loops` argument of `igraph_weighted_adjacency()` was converted to an `igraph_loops_t` for sake of consistency with `igraph_adjacency()` and `igraph_get_adjacency()`.
 - `igraph_write_graph_gml()` takes an additional bitfield parameter controlling some aspects of writing the GML file.
 - The `add_edges()` function in the attribute handler now takes an `igraph_vector_int_t` for its `edges` parameter instead of an `igraph_vector_t`. The `add_vertices()` function now takes an `igraph_integer_t` for the vertex count instead of a `long int`. The `combine_vertices()` and `combine_edges()` functions now take an `igraph_vector_ptr_t` containing vectors of type `igraph_vector_int_t` in their `merges` parameters. The `get_info()` function now uses `igraph_vector_int_t` to return the types of the graph, vertex and edge attribute types. The `permute_vertices()` and `permute_edges()` functions in the attribute handler tables now take an `igraph_vector_int_t` instead of an `igraph_vector_t` for the index vectors. These are relevant only to maintainers of higher level interfaces to igraph; they should update their attribute handlers accordingly.
 - igraph functions that interface with external libraries such as BLAS or LAPACK may now fail if the underlying BLAS or LAPACK implementation cannot handle the size of input vectors or matrices (BLAS and LAPACK are usually limited to vectors whose size fits in an `int`). `igraph_blas_dgemv()` and `igraph_blas_dgemv_array()` thus now return an `igraph_error_t`, which may be set to `IGRAPH_EOVERFLOW` if the input vectors or matrices are too large.
 - Functions that used an `igraph_vector_t` to represent cluster size and cluster membership now use an `igraph_vector_int_t` instead. These are:
   - `igraph_connected_components()` (used to be `igraph_clusters()` in 0.9 and before)
   - `igraph_community_eb_get_merges()`
   - `igraph_community_edge_betweenness()`
   - `igraph_community_fastgreedy()`
   - `igraph_community_fluid_communities()`
   - `igraph_community_infomap()`
   - `igraph_community_label_propagation()`
   - `igraph_community_leading_eigenvector()`
   - `igraph_community_leiden()`
   - `igraph_community_multilevel()`
   - `igraph_community_optimal_modularity()`
   - `igraph_community_spinglass()`
   - `igraph_community_spinglass_single()`
   - `igraph_community_to_membership()`
   - `igraph_community_walktrap()`
   - `igraph_compare_communities()`
   - `igraph_le_community_to_membership()`
   - `igraph_modularity()`
   - `igraph_reindex_membership()`
   - `igraph_split_join_distance()`
   - `igraph_community_multilevel()` additionally uses a `igraph_matrix_int_t` instead of `igraph_matrix_t()` for its memberships parameter.
 - `IGRAPH_TOTAL` was removed from the `igraph_neimode_t` enum; use the equivalent `IGRAPH_ALL` instead.

### Added

 - A new integer type, `igraph_uint_t` has been added. This is the unsigned pair of `igraph_integer_t` and they are always consistent in size.
 - A new container type, `igraph_vector_list_t` has been added, replacing most uses of `igraph_vector_ptr_t` in the API where it was used to hold a variable-length list of vectors. The type contains `igraph_vector_t` objects, and it is fully memory managed (i.e. its contents do not need to be allocated and destroyed manually). There is also a variant named `igraph_vector_int_list_t` for vectors of `igraph_vector_int_t` objects.
 - A new container type, `igraph_matrix_list_t` has been added, replacing most uses of `igraph_vector_ptr_t` in the API where it was used to hold a variable-length list of matrices. The type contains `igraph_matrix_t` objects, and it is fully memory managed (i.e. its contents do not need to be allocated and destroyed manually).
 - A new container type, `igraph_graph_list_t` has been added, replacing most uses of `igraph_vector_ptr_t` in the API where it was used to hold a variable-length list of graphs. The type contains `igraph_t` objects, and it is fully memory managed (i.e. its contents do not need to be allocated and destroyed manually).
 - The vector container type, `igraph_vector_t`, has been extended with a new variant whose functions all start with `igraph_vector_fortran_int_...`. This vector container can be used for interfacing with Fortran code as it guarantees that the integers in the vector are compatible with Fortran integers. Note that `igraph_vector_int_t` is not suitable any more, as the elements of `igraph_vector_int_t` are of type `igraph_integer_t`, whose size may differ on 32-bit and 64-bit platforms, depending on how igraph was compiled.
 - `igraph_adjlist_init_from_inclist()` to create an adjacency list from an already existing incidence list by resolving edge IDs to their corresponding endpoints. This function is useful for algorithms when both an adjacency and an incidence list is needed and they should be in the same order.
 - `igraph_almost_equals()` and `igraph_cmp_epsilon()` to compare floating point numbers with a relative tolerance.
 - `igraph_betweenness_subset()` and `igraph_edge_betweenness_subset()` calculates betweenness and edge betweenness scores using shortest paths between a subset of vertices only (#1711, thanks to @guyroznb)
 - `igraph_blas_dgemm()` to multiply two matrices.
 - `igraph_calloc()` and `igraph_realloc()` are now publicly exposed; these functions provide variants of `calloc()` and `realloc()` that can safely be deallocated within igraph functions.
 - `igraph_circulant()` to create circulant graphs (#1856, thanks to @Gomango999).
 - `igraph_complex_almost_equals()` to compare complex numbers with a relative tolerance.
 - `igraph_eccentricity_dijkstra()` finds the longest weighted path length among all shortest paths between a set of vertices.
 - `igraph_enter_safelocale()` and `igraph_exit_safelocale()` for temporarily setting the locale to C. Foreign format readers and writers require a locale which uses a decimal point instead of decimal comma.
 - `igraph_es_all_between()` to create an edge selector that selects all edges between a pair of vertices.
 - `igraph_full_multipartite()` generates full multipartite graphs (a generalization of bipartite graphs to multiple groups).
 - `igraph_fundamental_cycles()` computes a fundamental cycle basis (experimental).
 - `igraph_generalized_petersen()` to create generalized Petersen graphs (#1844, thanks to @alexsyou).
 - `igraph_get_all_eids_between()` returns the IDs of all edges between a pair of vertices.
 - `igraph_get_k_shortest_paths()` finds the k shortest paths between a source and a target vertex.
 - `igraph_get_laplacian()` and `igraph_get_laplacian_sparse()` return the Laplacian matrix of the graph as a dense or sparse matrix, with various kinds of normalizations. They replace the now-deprecated `igraph_laplacian()` function. This makes the API consistent with `igraph_get_adjacency()` and `igraph_get_adjacency_sparse()`.
 - `igraph_get_widest_path()`, `igraph_get_widest_paths()`, `igraph_widest_path_widths_dijkstra()` and `igraph_widest_path_widths_floyd_warshall()` to find widest paths (#1893, thanks to @Gomango999).
 - `igraph_graph_center()` finds the central vertices of the graph. The central vertices are the ones having a minimum eccentricity (PR #2084, thanks to @pradkrish).
 - `igraph_graph_count()` returns the number of unlabelled graphs on a given number of vertices. It is meant to find the maximum isoclass value.
 - `igraph_has_mutual()` checks if a directed graph has any mutual edges.
 - `igraph_heap_clear()` and `igraph_heap_min_clear()` remove all elements from an `igraph_heap_t` or an `igraph_heap_min_t`, respectively.
 - `igraph_invalidate_cache()` invalidates all cached graph properties, forcing their recomputation next time they are requested. This function should not be needed in everyday usage, but may be useful in debugging and benchmarking.
 - `igraph_is_forest()` to check whether a graph is a forest (#1888, thanks to @rohitt28).
 - `igraph_is_acyclic()` to check whether a graph is acyclic (#1945, thanks to @borsgeorgica).
 - `igraph_is_perfect()` to check whether a graph is a perfect graph (#1730, thanks to @guyroznb).
 - `igraph_hub_and_authority_scores()` calculates the hub and authority scores of a graph as a matching pair.
 - `igraph_layout_umap()` and `igraph_layout_umap_3d()` to lay out a graph in 2D or 3D space using the UMAP dimensionality reduction algorithm.
 - `igraph_local_scan_subset_ecount()` counts the number of edges in induced sugraphs from a subset of vertices.
 - `igraph_matrix_view_from_vector()` allows interpreting the data stored in a vector as a matrix of the specified size.
 - `igraph_minimum_cycle_basis()` computes an unweighted minimum cycle basis (experimental).
 - `igraph_pseudo_diameter()` and `igraph_pseudo_diameter_dijkstra()` to determine a lower bound for the diameter of a graph (unweighted or weighted).
 - `igraph_regular_tree()` creates a regular tree where all internal vertices have the same total degree.
 - `igraph_rngtype_pcg32` and `igraph_rngtype_pcg64` implement 32-bit and 64-bit variants of the PCG random number generator.
 - `igraph_rng_get_pois()` generates random variates from the Poisson distribution.
 - `igraph_roots_for_tree_layout()` computes a set of roots suitable for a nice tree layout.
 - `igraph_spanner()` calculates a spanner of a graph with a given stretch factor (#1752, thanks to @guyroznb)
 - `igraph_sparse_adjacency()` and `igraph_sparse_weighted_adjacency()` constructs graphs from (weighted) sparse matrices.
 - `igraph_sparsemat_get()` to retrieve a single element of a sparse matrix.
 - `igraph_sparsemat_normalize_rows()` and `igraph_sparsemat_normalize_cols()` to normalize sparse matrices row-wise or column-wise.
 - `igraph_stack_capacity()` to query the capacity of a stack.
 - `igraph_strvector_capacity()` returns the maximum number of strings that can be stored in a string vector without reallocating the memory block holding the pointers to the individual strings.
 - `igraph_strvector_merge()` moves all strings from one string vectors to the end of another without re-allocating them.
 - `igraph_strvector_push_back_len()` adds a new string to the end of a string vector and allows the user to specify the length of the string being added.
 - `igraph_strvector_reserve()` reserves space for a given number of string pointers in a string vector.
 - `igraph_symmetric_tree()` to create a tree with the specified number of branches at each level (#1859, thanks to @YuliYudith and @DoruntinaM).
 - `igraph_trussness()` calculates the trussness of each edge in the graph (#1034, thanks to @alexperrone)
 - `igraph_turan()` generates Turán graphs (#2088, thanks to @pradkrish)
 - `igraph_vector_all_almost_e()`, `igraph_vector_complex_all_almost_e()`, `igraph_matrix_all_almost_e()`, `igraph_matrix_complex_all_almost_e()` for elementwise comparisons of floating point vector and matrices with a relative tolerance.
 - `igraph_vector_complex_zapsmall()` and `igraph_matrix_complex_zapsmall()` for replacing small components of complex vector or matrix elements with exact zeros.
 - `igraph_vector_lex_cmp_untyped()` and `igraph_vector_colex_cmp_untyped()` for lexicographic and colexicographic comparison of vectors, similarly to `igraph_vector_lex_cmp()` and `igraph_vector_colex_cmp()`. The difference between the two variants is that the untyped versions declare the vectors as `const void*`, making the functions suitable as comparators for `qsort()`.
 - `igraph_vector_permute()` functions to permute a vector based on an index vector.
 - `igraph_vector_ptr_sort_ind()` to obtain an index vector that would sort a vector of pointers based on some comparison function.
 - `igraph_vector_range()` to fill an existing vector with a range of increasing numbers.
 - `igraph_vector_remove_fast()` functions to remove an item from a vector by swapping it with the last element and then popping it off. It allows one to remove an item from a vector in constant time if the order of items does not matter.
 - `igraph_vertex_path_from_edge_path()` converts a sequence of edge IDs representing a path to an equivalent sequence of vertex IDs that represent the vertices the path travelled through.
 - `igraph_vs_range()`, `igraph_vss_range()`, `igraph_es_range()` and `igraph_ess_range()` creates vertex and edge sequences from C-style intervals (closed from the left, open from the right).
 - `igraph_wheel()` to create a wheel graph (#1938, thanks to @kwofach).

### Removed

 - `igraph_adjlist_remove_duplicate()`, `igraph_betweenness_estimate()`, `igraph_closeness_estimate()`, `igraph_edge_betweenness_estimate()`, `igraph_inclist_remove_duplicate()`, `igraph_is_degree_sequence()` and `igraph_is_graphical_degree_sequence()` were deprecated earlier in 0.9.0 and are now removed in this release.
 - `igraph_dnorm()`, `igraph_strvector_move_interval()`, `igraph_strvector_permdelete()` and `igraph_strvector_remove_negidx()` were removed. These are not breaking changes as the functions were never documented, they were only exposed from one of the headers.
 - `igraph_eigen_laplacian()`, `igraph_es_fromto()` and `igraph_maximum_matching()` were removed. These are not breaking changes either as the functions were never implemented, they returned an error code unconditionally.

### Changed

 - `igraph_degree_sequence_game()` now supports an additional method, `IGRAPH_DEGSEQ_EDGE_SWITCHING_SIMPLE`, an edge-switching MCMC sampler.
 - `igraph_get_adjacency()` and `igraph_get_adjacency_sparse()` now count loop edges _twice_ in undirected graphs when using `IGRAPH_GET_ADJACENCY_BOTH`. This is to ensure consistency with `IGRAPH_GET_ADJACENCY_UPPER` and `IGRAPH_GET_ADJACENCY_LOWER` such that the sum of the upper and the lower triangle matrix is equal to the full adjacency matrix even in the presence of loop edges.
 - `igraph_matrix_print()` and `igraph_matrix_fprint()` functions now align columns when priting.
 - `igraph_read_graph_gml()` now supports graph attributes (in addition to vertex and edge attributes).
 - `igraph_read_graph_gml()` now uses NaN as the default numerical attribute values instead of 0.
 - The Pajek parser in `igraph_read_graph_pajek()` is now less strict and accepts more files.
 - `igraph_ring()` no longer simplifies its result when generating a one- or two-vertex graph. The one-cycle has a self-loop and the undirected two-cycle has parallel edges.
 - `igraph_vector_view()` now allows `data` to be `NULL` in the special case when `length == 0`.
 - `igraph_version()` no longer returns an error code.
 - `igraph_write_graph_gml()` uses the `creator` parameter in a different way: the supplied string is now written into the Creator line as-is instead of being appended to a default value.
 - `igraph_write_graph_gml()` skips writing NaN values. These two changes ensure consistent round-tripping.
 - `igraph_write_graph_gml()` and `igraph_read_graph_gml()` now have limited support for entity encoding.
 - `igraph_write_graph_ncol()` now preserves the edge ordering of the graph when writing an NCOL file.
 - igraph functions that take an ARPACK options object now also accept `NULL` in place of an options object, and they will fall back to using a default object provided by `igraph_arpack_options_get_default()`.
 - Foreign format readers now present more informative error messages.
 - The default tolerance of the zapsmall functions is now `eps^(2/3)` instead of `eps^(1/2)` where eps is the machine epsilon of `igraph_real_t`.
 - It is now possible to override the uniform integer and the Poisson samplers in the random number generator interface.

### Fixed

 - When an error occurs during parsing DL, GML, NCOL, LGL or Pajek files, line numbers are now reported correctly.
 - The GraphML parser does not print to stderr any more in case of encoding errors and other error conditions originating from the underlying `libxml2` library.
 - The GraphML parser would omit some edges and vertices when reading files with custom attribute types, such as those produced by yEd. This is now corrected.
 - The GML parser no longer mixes up Inf and NaN and -Inf now works.
 - The GML parser now supports nodes with no id field.
 - The GML parser now performs more stringent checks on the input file, such as verifying that `id`, `source`, `target` and `directed` fields are not duplicated.
 - The core data structures (vector, etc.) have overflow checks now.
 - Deterministic graph generators, as well as most random ones, have overflow checks now.
 - Graphs no longer lose all their attributes after calling `igraph_contract_vertices()`.
 - `igraph_hrg_init()` does not throw an assertion error anymore for zero vertices.
 - `igraph_matrix_complex_create()` and `igraph_matrix_complex_create_polar()` now set their sizes correctly.
 - `igraph_random_walk()` took one fewer steps than specified.
 - `igraph_sparsemat_getelements_sorted()` did not sort the elements for triplet matrices correctly; this is fixed now.
 - `igraph_write_graph_gml()` no longer produces corrupt output when some string attribute values contain `"` characters.

### Deprecated

 - `igraph_clusters()` has been renamed to `igraph_connected_components()`; the old name is deprecated and will be removed in 0.11.
 - `igraph_complex_eq_tol()` is now deprecated in favour of `igraph_complex_almost_equals()`.
 - `igraph_get_sparsemat()` is deprecated in favour of `igraph_get_adjacency_sparse()`, and will be removed in 0.11. Note that `igraph_get_adjacency_sparse()` takes an _initialized_ sparse matrix as input, unlike `igraph_get_sparsemat()` which takes an uninitialized one.
 - `igraph_get_stochastic_sparsemat()` is deprecated in favour of `igraph_get_stochastic_sparse()`, and will be removed in 0.11. Note that `igraph_get_stochastic_sparse()` takes an _initialized_ sparse matrix as input, unlike `igraph_get_stochastic_sparsemat()`, which takes an uninitialized one.
 - `igraph_isomorphic_34()` has been deprecated in favour of `igraph_isomorphic()`.  Note that `igraph_isomorphic()` calls an optimized version for directed graphs of size 3 and 4, and undirected graphs with 3-6 vertices, so there is no need for a separate function.
 - `igraph_laplacian()` is now deprecated; use `igraph_get_laplacian()` or `igraph_get_laplacian_sparse()` depending on whether you need a dense or a sparse matrix.
 - `igraph_lattice()` has been renamed to `igraph_square_lattice()` to indicate that this function generates square lattices only. The old name is deprecated and will either be removed in 0.11 or will be changed to become a generic lattice generator that also supports other types of lattices.
 - `igraph_local_scan_neighborhood_ecount()` is now deprecated in favour of `igraph_local_scan_subset_ecount()`.
 - `igraph_matrix_all_e_tol()` is now deprecated in favour of `igraph_matrix_all_almost_e()`.
 - `igraph_matrix_copy()` is now deprecated; use `igraph_matrix_init_copy()` instead. The new name emphasizes that the function _initializes_ the first argument instead of expecting an already-initialized target matrix. The old name will be removed in 0.11.
 - `igraph_matrix_e()` and `igraph_matrix_e_ptr()` have been renamed to `igraph_matrix_get()` and `igraph_matrix_get_ptr()`. The old names are deprecated and will be removed in 0.11.
- `igraph_random_edge_walk()` has been deprecated by `igraph_random_walk()` to support edges and/or vertices for the random walk in a single function.  It will be removed in 0.11.
 - `igraph_read_graph_dimacs()` has been renamed to `igraph_read_graph_dimacs_flow()`; the old name is deprecated and might be re-used as a generic DIMACS reader in the future. Also, the function now uses `igraph_integer_t` as the source and target vertex IDs instead of a `long int`.
 - `igraph_shortest_paths()` and related functions were renamed to `igraph_distances()`; the old name was unfortunate because these functions calculated _path lengths_ only and not the paths themselves. The old names are deprecated and will be removed in 0.11.
 - `igraph_sparsemat_copy()`, `igraph_sparsemat_diag()` and `igraph_sparsemat_eye()` have been renamed to `igraph_sparsemat_init_copy()`, `igraph_sparsemat_init_diag()` and `igraph_sparsemat_init_eye()` to indicate that they _initialize_ a new sparse matrix. The old names are deprecated and will be removed in 0.11.
 - `igraph_strvector_add()` has been renamed to `igraph_strvector_push_back()` for sake of consistency with other vector-like data structures; the old name is deprecated and will be removed in 0.11.
 - `igraph_strvector_copy()` has been renamed to `igraph_strvector_init_copy()` for sake of consistency with other vector-like data structures; the old name is deprecated and will be removed in 0.11.
 - `igraph_strvector_get()` now returns a `const char*` and not a `char*` to indicate that you are not supposed to modify the string in the vector directly. If you do want to modify it and you are aware of the implications (i.e. the new string must not be longer than the original one), you can cast away the constness of the return value before modifying it.
 - `igraph_strvector_set2()` has been renamed to `igraph_strvector_set_len()`; the old name is deprecated and will be removed in 0.11.
 - `igraph_tree()` has been renamed to `igraph_kary_tree()`; the old name is deprecated and will be removed in 0.11.
 - `igraph_vector_e()` and `igraph_vector_e_ptr()` have been renamed to `igraph_vector_get()` and `igraph_vector_get_ptr()`. The old names are deprecated and will be removed in 0.11.
 - `igraph_vector_e_tol()` is now deprecated in favour of `igraph_vector_all_almost_e()`.
 - `igraph_vector_copy()` is now deprecated; use `igraph_vector_init_copy()` instead. The new name emphasizes that the function _initializes_ the first argument instead of expecting an already-initialized target vector. The old name will be removed in 0.11.
 - `igraph_vector_init_seq()` is now deprecated in favour of `igraph_vector_init_range()`, which uses C-style intervals (closed from the left and open from the right).
 - `igraph_vs_seq()`, `igraph_vss_seq()`, `igraph_es_seq()` and `igraph_ess_seq()` are now deprecated in favour of `igraph_vs_range()`, `igraph_vss_range()`, `igraph_es_range()` and `igraph_ess_range()` because these use C-style intervals (closed from the left, open from the right).
 - `igraph_write_graph_dimacs()` has been renamed to `igraph_write_graph_dimacs_flow()`; the old name is deprecated and might be re-used as a generic DIMACS writer in the future. Also, the function now uses `igraph_integer_t` as the source and target vertex IDs instead of a `long int`.
 - `igraph_zeroin()` is deprecated and will be removed in 0.11, with no replacement. The function is not graph-related and was never part of the public API.
 - The macros `igraph_Calloc`, `igraph_Realloc` and `igraph_Free` have been deprecated in favour of `IGRAPH_CALLOC`, `IGRAPH_REALLOC` and `IGRAPH_FREE` to simplify the API. The deprecated variants will be removed in 0.11.

### Other

 - Documentation improvements.
 - Support for Intel's LLVM-based compiler.

## [0.9.10] - 2022-09-02

### Added

 - `igraph_reverse_edges()` reverses the specified edges in the graph while preserving all attributes.

### Changed

 - The `IGRAPH_ARPACK_PROD` error code is no longer used. Instead, the specific error encountered while doing matrix multiplication is reported.
 - XML external entities are not resolved any more when parsing GraphML files to prevent XML external entity injection (XXE) attacks. Standard XML entities like `&lt;` or `&quot;` still work.

### Fixed

 - Fixed incorrect results from `igraph_local_scan_1_ecount()` when the graph was directed but the mode was `IGRAPH_ALL` and some nodes had loop edges. See issue #2092.
 - Fixed incorrect counting of self-loops in `igraph_local_scan_neighborhood_ecount()` when the graph was undirected.
 - In some rare edge cases, `igraph_pagerank()` with the ARPACK method and `igraph_hub_score()` / `igraph_authority_score()` could return incorrect results. The problem could be detected by checking that the returned eigenvalue is not negative. See issue #2090.
 - `igraph_permute_vertices()` now checks for out-of-range indices and duplicates in the permutation vector.
 - `igraph_create()` now checks for non-finite vertex indices in the edges vector.
 - `igraph_eigenvector_centrality()` would return incorrect scores when some weights were negative.
 - `igraph_es_seq()` and `igraph_ess_seq()` did not include the `to` vertex in the sequence.
 - `igraph_eit_create()` and `igraph_vit_create()` now check that all edge/vertex indices are in range when creating iterators from sequence-type selectors.
 - `igraph_grg_game()` now validates its arguments.
 - `igraph_layout_drl()` and its 3D version now validate their inputs.
 - `igraph_layout_kamada_kawai()`, `igraph_layout_fruchterman_reingold()`, `igraph_layout_drl()`, as well as their 3D versions now check for non-positive weights.
 - `igraph_asymmetric_preference_game()` interpreted its `type_dist_matrix` argument incorrectly.
 - Fixed incorrect result of `igraph_community_spinglass()` for null and singleton graphs.
 - `igraph_layout_gem()` does not crash any more for graphs with only a single vertex.
 - `igraph_bridges()` no longer uses recursion and thus is no longer prone to stack overflow.
 - Include paths of dependent packages would be specified incorrectly in some environments.

### Other

 - Documentation improvements.

## [0.9.9] - 2022-06-04

### Changed

 - `igraph_community_walktrap()` now uses double precision floating point operations internally instead of single precision.
 - In `igraph_community_leiden()`, the `nb_clusters` output parameter is now optional (i.e. it can be `NULL`).
 - `igraph_read_graph_graphml()` no longer attempts to temporarily set the C locale, and will therefore not work correctly if the current locale uses a decimal comma.

### Fixed

 - `igraph_community_walktrap()` would return an invalid `modularity` vector when the `merges` matrix was not requested.
 - `igraph_community_walktrap()` would return a `modularity` vector that was too long for disconnected graphs. This would cause a failure in some weighted graphs when the `membership` vector was requested.
 - `igraph_community_walktrap()` now checks the weight vector: only non-negative weights are accepted, and all vertices must have non-zero strength.
 - `igraph_community_walktrap()` now returns a modularity score of NaN for graphs with no edges.
 - `igraph_community_fast_greedy()` now returns a modularity score of NaN for graphs with no edges.
 - `igraph_community_edge_betweenness()` now returns a modularity vector with a single NaN entry for graph with no edges. Previously it returned a zero-length vector.
 - `igraph_community_leading_eigenvector()` does not ignore non-ARPACK-related errors from `igraph_arpack_rssolve()` any more.
 - `igraph_preference_game()` now works correctly when `fixed_size` is true and
   `type_dist` is not given; earlier versions had a bug where more than half of
   the vertices mistakenly ended up in group 0.
 - Fixed a memory leak in `igraph_hrg_fit()` when using `start=1`.
 - `igraph_write_graph_dot()` now outputs NaN values unchanged.
 - `igraph_write_graph_dot()` no longer produces invalid DOT files when empty string attributes are present.
 - `igraph_layout_fruchterman_reingold()` and `igraph_layout_kamada_kawai()`, as well as their 3D versions, did not respect vertex coordinate bounds (`xmin`, `xmax`, etc.) when minimum values were large or maximum values were small. This is now fixed.
 - The initial coordinates of the Kamada-Kawai layout (`igraph_layout_kamada_kawai()` and `igraph_layout_kamada_kawai_3d()`) are chosen to be more in line with the original publication, improving the stability of the result. See isse #963. This changes the output of the function for the same graph, compared with previous versions. To obtain the same layout, initialize coordinates with `igraph_layout_circle()` (in 2D) or `igraph_layout_sphere()` (in 3D).
 - Improved numerical stability in Kamada-Kawai layout.
 - Corrected a problem in the calculation of displacements in `igraph_layout_fruchterman_reingold()` and its 3D version. This fixes using the "grid" variant of the algorithm on disconnected graphs.
 - `igraph_sumtree_search()` would consider search intervals open on the left and closed on the right, contrary to the documentation. This is now corrected to closed on the left and open on the right. In some cases this lead to a zero-weight element being returned for a zero search value. See issue #2080.

### Other

 - Greatly improved error reporting from foregin format parsers.
 - Documentation improvements.

## [0.9.8] - 2022-04-08

### Fixed

 - Assertion failure in `igraph_bfs()` when an empty `roots` or `restricted` vector was provided.
 - `igraph_diversity()` now returns 0 for degree-1 vertices. Previously it incorrectly returned NaN or +-Inf depending on roundoff errors.
 - `igraph_community_walktrap()` does not crash any more when provided with
   `modularity=NULL` and `membership=NULL`.

### Other

 - Documentation improvements.

## [0.9.7] - 2022-03-16

### Changed

 - `igraph_get_all_shortest_paths_dijsktra()` now uses tolerances when comparing path
   lengths, and is thus robust to numerical roundoff errors.
 - `igraph_vector_*_swap` and `igraph_matrix_swap` now take O(1) instead of O(n) and accept all sizes.

### Fixed

 - NCOL and LGL format writers no longer accept "name" and "weight" attributes
   of invalid types.
 - The LGL writer could not access numerical weight attributes, potentially leading
   to crashes.
 - External PLFIT libraries and their headers are now detected at their standard
   installation location.
 - `igraph_vector_init()` no longer accepts negative vector sizes.
 - `igraph_assortativity_nominal()` crashed on the null graph.
 - Label propagation now ensures that all labels are dominant.
 - Fixed incorrect partition results for walktrap algorithm (issue #1927)
 - Negative values returned by `igraph_rng_get_integer()` and `RNG_INTEGER()` were incorrect,
   one larger than they should have been.
 - `igraph_community_walktrap()` now checks its `steps` input argument.
 - The first modularity value reported by `igraph_community_walktrap()` was
   incorrect (it was always zero). This is now fixed.
 - `igraph_correlated_game()` would return incorrect results, or exhaust the memory,
    for most input graphs that were not generated with `igraph_erdos_renyi_game_gnp()`.
 - `igraph_community_label_propagation` incorrectly did not result in all labels being dominant (issue #1963, fixed in PR #1966).

### Other

 - The C attribute handler now verifies attribute types when retrieving attributes.
 - Documentation improvements.

## [0.9.6] - 2022-01-05

 - Isomorphism class functions (`igraph_isoclass()`, `igraph_isoclass_subgraph()`,
   `igraph_isoclass_create`) and motif finder functions (`igraph_motifs_randesu()`,
   `igraph_motifs_randesu_estimate()`, `igraph_motifs_randesu_callback()`) now
   support undirected (sub)graphs of sizes 5 and 6. Previsouly only sizes 3 and 4
   were supported.

### Fixed

 - igraph would not build with MinGW when using the vendored GLPK and enabling TLS.
 - Removed some uses of `abort()` from vendored libraries, which could unexpectedly
   shut down the host language of igraph's high-level interfaces.
 - `igraph_community_label_propagation()` no longer leaves any vertices unlabeled
   when they were not reachable from any labeled ones, i.e. the returned membership
   vector is guaranteed not to contain negative values (#1853).
 - The Kamada-Kawai layout is now interruptible.
 - The Fruchterman-Reingold layout is now interruptible.
 - Fixed a bug in `igraph_cmp_epsilon()` that resulted in incorrect results for
   edge betweenness calculations in certain rare cases with x87 floating point
   math when LTO was also enabled (#1894).
 - Weighted clique related functions now fall back to the unweighted variants
   when a null vertex weight vector is given to them.
 - `igraph_erdos_renyi_game_(gnm|gnp)` would not produce self-loops for the singleton
   graph.
 - Fixed a bug in `igraph_local_efficiency()` that sometimes erroneously
   reported zero as the local efficiency of a vertex in directed graphs.
 - `igraph_vector_update()` (and its type-specific variants) did not check for
   memory allocation failure.
 - Fixed a potential crash in the GraphML reader that would be triggered by some
   invalid GraphML files.

### Other

 - `igraph_is_tree()` has improved performance and memory usage.
 - `igraph_is_connected()` has improved performance when checking weak connectedness.
 - Improved error handling in `igraph_maximal_cliques()` and related functions.
 - The build system now checks that GLPK is of a compatible version (4.57 or later).
 - The vendored `plfit` package was updated to 0.9.3.
 - You can now build igraph with an external `plfit` instead of the vendored one.
 - Documentation improvements.

## [0.9.5] - 2021-11-11

### Fixed

 - `igraph_reindex_membership()` does not allow negative membership indices any more.

 - `igraph_rewire_directed_edges()` now generates multigraphs when edge directions
   are ignored, to make it consistent with the directed case.

 - Fixed a bug in `igraph_gomory_hu_tree()` that returned only the equivalent flow
   tree instead of the cut tree (#1810).

 - Fixed a bug in the `IGRAPH_TO_UNDIRECTED_COLLAPSE` mode of
   `igraph_to_undirected()` that provided an incorrect merge vector to the
   attribute handler, leading to problems when edge attributes were merged
   using an attribute combination (#1814).

 - Fixed the behaviour of the `IGRAPH_ENABLE_LTO` option when it was set to
   `AUTO`; earlier versions had a bug where `AUTO` simply checked whether LTO
   is supported but then did not use LTO even if it was supported.

 - When using igraph from a CMake project, it is now checked that the project has
   the C++ language enabled. This is necessary for linking to igraph with CMake.

### Other

 - Improved the root selection method for disconnected graphs in the
   Reingold-Tilford layout (#1836). The new root selection method provides
   niceer results if the graph is not a tree, although it is still recommended
   to use the Sugiyama layout instead, unless the input graph is _almost_ a
   tree, in which case Reingold-Tilfold may still be preferred.

 - `igraph_decompose()` is now much faster for large graphs containing many
   isolates or small components (#960).

 - `igraph_largest_cliques()` and `igraph_clique_number()` were re-written to
   use `igraph_maximal_cliques_callback()` so they are much faster now (#804).

 - The vendored GLPK has been upgraded to GLPK 5.0.

 - Documentation improvements.

## [0.9.4] - 2021-05-31

### Changed

 - Unweighted transitivity (i.e. clustering coefficient) calculations now ignore multi-edges and edge directions instead of rejecting multigraphs and directed graphs.
 - `igraph_transitivity_barrat()` now returns an error code if the input graph has multiple edges (which is not handled correctly by the implementation yet).

### Fixed

 - `igraph_local_scan_k_ecount()` now handles loops correctly.
 - `igraph_transitivity_avglocal_undirected()` is no longer slower than `igraph_transitivity_local_undirected()`.
 - Worked around an invalid warning issued by Clang 9.0 when compiling with OpenMP.

### Other

 - Documentation improvements.

## [0.9.3] - 2021-05-05

### Added

 - OpenMP is now enabled and used by certain functions (notably PageRank calculation) when the compiler supports it. Set `IGRAPH_OPENMP_SUPPORT=OFF` at configuration time to disable this.

### Fixed

 - `igraph_get_incidence()` no longer reads and writes out of bounds when given a non-bipartite graph, but gives a warning and ignores edges within a part.
 - `igraph_dyad_census()` no longer reports an overflow on singleton graphs, and handles loops and multigraphs correctly. Undirected graphs are handled consistently and will no longer give a warning.
 - `igraph_vector_lex_cmp()` and `igraph_vector_colex_cmp()` dereferenced their arguments only once instead of twice, and therefore did not work with `igraph_vector_ptr_sort()`.
 - `igraph_maximal_cliques_subset()` and `igraph_transitivity_barrat()` corrupted the error handling stack ("finally stack") under some circumstances.
 - CMake package files did not respect `CMAKE_INSTALL_LIBDIR`. This only affected Linux distributions which install into `lib64` or other locations instead of `lib`.
 - The parser sources could not be generated when igraph was in a location that contained spaces in its path.
 - igraph no longer links to the math library (`libm`) when this is not necessary.
 - `_CRT_SECURE_NO_WARNINGS` is now defined during compilation to enable compatibility with UWP.
 - Fixed a compilation issue on MSYS / MinGW when link-time optimization was enabled and the `MSYS Makefiles` CMake generator was used. Some source files in igraph were renamed as a consequence, but these should not affect users of the library.

### Deprecated

 - `igraph_rng_min()` is now deprecated; assume a constant zero as its return value if you used this function in your own code.

### Other

 - Updated the vendored CXSparse library to version 3.2.0

## [0.9.2] - 2021-04-14

### Added

 - CMake package files are now installed with igraph. This allows `find_package(igraph)` to find igraph and detect the appropriate compilation options for projects that link to it.

### Fixed

 - igraph can now be used as a CMake subproject in other CMake-based projects.
 - The documentaton can now be built from the release tarball.
 - Configuration will no longer fail when the release tarball is extracted into a subdirectory of an unrelated git repository.
 - The generated pkg-config file was incorrect when `CMAKE_INSTALL_<dir>` variables were absolute paths.
 - On Unix-like systems, the library name is now `libigraph.so.0.0.0`, as it used to be for igraph 0.8 and earlier.
 - Fixed a return type mismatch in parser sources, and fixed some warnings with recent versions of gcc.
 - Fixed a bug in `igraph_get_shortest_paths_dijkstra()` and `igraph_get_shortest_paths_bellman_ford()` that returned incorrect results for unreachable vertices.

### Other

 - Improved installation instructions and tutorial.

## [0.9.1] - 2021-03-23

### Added

 - `igraph_vector_lex_cmp()` and `igraph_vector_colex_cmp()` for lexicographic
   and colexicographic comparison of vectors. These functions may also be used
   for sorting.

### Changed

 - `igraph_community_multilevel()` is now randomized (PR #1696, thanks to Daniel Noom).

### Fixed

 - CMake settings that controlled the library installation directory name, such as `CMAKE_INSTALL_LIBDIR`, were not respected.
 - Under some conditions, the generated pkg-config file contained an incorrect include directory path.
 - The following functions were not exported from the shared library: `igraph_subcomponent()`, `igraph_stack_ptr_free_all()`, `igraph_stack_ptr_destroy_all()`, `igraph_status_handler_stderr()`, `igraph_progress_handler_stderr()`.
 - Built-in random number generators (`igraph_rngtype_mt19937`, `igraph_rngtype_rand`, `igraph_rngtype_glibc2`) were not exported from the shared library.
 - `igraph_layout_graphopt()` no longer rounds the `spring_length` parameter to an integer.
 - `igraph_get_all_shortest_paths_dijkstra()` no longer modifies the `res` vector's item destructor.
 - `igraph_get_shortest_path_bellman_ford()` did not work correctly when calculating paths to all vertices.
 - `igraph_arpack_rnsolve()` checks its parameters more carefully.
 - `igraph_community_to_membership()` does not crash anymore when `csize` is requested but `membership` is not.
 - `igraph_citing_cited_type_game()`: fixed memory leaks (PR #1700, thanks to Daniel Noom).
 - `igraph_transitivity_undirected()`, `igraph_transitivity_avglocal_undirected()` and `igraph_transitivity_barrat()` no longer trigger an assertion failure when used with the null graph (PRs #1709, #1710).
 - `igraph_(personalized_)pagerank()` would return incorrect results for weighted multigraphs with fewer than 128 vertices when using `IGRAPH_PAGERANK_ALGO_PRPACK`.
 - `igraph_diversity()` now checks its input more carefully, and throws an error when the input graph has multi-edges or is directed.
 - `igraph_shortest_paths_johnson()` would return incorrect results when the `to` argument differed from `from` (thanks to Daniel Noom).
 - `igraph_is_graphical()` would fail to set the result variable for certain special degree sequences in the undirected simple graph case.
 - Non-maximal clique finding functions would sometimes return incomplete results when finding more than 2147483647 (i.e. 2^31 - 1) cliques.
 - GLPK internal errors no longer crash igraph.
 - Fixed some potential memory leaks that could happen on error conditions or when certain functions were interrupted.
 - When testing a DLL build on Windows, the `PATH` was sometimes not set correctly, causing the tests to fail (PR #1692).
 - When compiling from the git repository (as opposed to the release tarball), the build would fail with recent versions of `bison` and `flex`.

### Other

 - Documentation improvements.
 - Much faster documentation builds.
 - Allow using a pre-generated `arith.h` header for f2c when cross-compiling; see the Installation section of the documentation.
 - The `IGRAPH_ENABLE_LTO` build option now supports the `AUTO` value, which uses LTO only if the compiler supports it. Warning: CMake may not always be able to detect that LTO is not fully supported. Therefore, the default setting is `OFF`.
 - The following functions are now interruptible: `igraph_grg_game()`, `igraph_sbm_game()`, `igraph_barabasi_game()`, `igraph_barabasi_aging_game()`.
 - Functions that use GLPK, such as `igraph_feedback_arc_set()` and `igraph_community_optimal_modularity()` are now interruptible.
 - Add support for older versions of Clang that do not recognize the `-Wno-varargs` flag.

### Acknowledgments

 - Big thanks to Daniel Noom for continuing to expand the test suite and discovering and fixing several bugs in the process!

## [0.9.0] - 2021-02-16

### Added

 - Eulerian paths/cycles (PR #1346):
   * `igraph_is_eulerian()` finds out whether an Eulerian path/cycle exists.
   * `igraph_eulerian_path()` returns an Eulerian path.
   * `igraph_eulerian_cycle()` returns an Eulerian cycle.
 - Efficiency (PR #1344):
   * `igraph_global_efficiency()` computes the global efficiency of a network.
   * `igraph_local_efficiency()` computes the local efficiency around each vertex.
   * `igraph_average_local_efficiency()` computes the mean local efficiency.
 - Degree sequences (PR #1445):
   * `igraph_is_graphical()` checks if a degree sequence has a realization as a simple or multigraph, with or without self-loops.
   * `igraph_is_bigraphical()` checks if two degree sequences have a realization as a bipartite graph.
   * `igraph_realize_degree_sequence()` now supports constructing non-simple graphs as well.
 - There is a new fatal error handling mechanism (PR #1548):
   * `igraph_set_fatal_handler()` sets the fatal error handler. It is the only function in this functionality group that is relevant to end users.
   * The macro `IGRAPH_FATAL()` and the functions `igraph_fatal()` and `igraph_fatalf()` raise a fatal error. These are for internal use.
   * `IGRAPH_ASSERT()` is a replacement for the `assert()` macro. It is for internal use.
   * `igraph_fatal_handler_abort()` is the default fatal error handler.
 - The new `IGRAPH_WARNINGF`, `IGRAPH_ERRORF` and `IGRAPH_FATALF` macros provide warning/error reporting with `printf`-like syntax. (PR #1627, thanks to Daniel Noom!)
 - `igraph_average_path_length_dijkstra()` computes the mean shortest path length in weighted graphs (PR #1344).
 - `igraph_get_shortest_paths_bellman_ford()` computes the shortest paths (including the vertex and edge IDs along the paths) using the Bellman-Ford algorithm (PR #1642, thanks to Guy Rozenberg). This makes it possible to calculate the shortest paths on graphs with negative edge weights, which was not possible before with Dijkstra's algorithm.
 - `igraph_get_shortest_path_bellman_ford()` is a wrapper for `igraph_get_shortest_paths_bellman_ford()` for the single path case.
 - `igraph_is_same_graph()` cheks that two labelled graphs are the same (PR #1604).
 - Harmonic centrality (PR #1583):
   * `igraph_harmonic_centrality()` computes the harmonic centrality of vertices.
   * `igraph_harmonic_centrality_cutoff()` computes the range-limited harmonic centrality.
 - Range-limited centralities, currently equivalent to the old functions with names ending in `_estimate` (PR #1583):
   * `igraph_closeness_cutoff()`.
   * `igraph_betweenness_cutoff()`.
   * `igraph_edge_betweenness_cutoff()`.
 - `igraph_vector_is_any_nan()` checks if any elements of an `igraph_vector_t` is NaN.
 - `igraph_inclist_size()` returns the number of vertices in an incidence list.
 - `igraph_lazy_adjlist_size()` returns the number of vertices in a lazy adjacency list.
 - `igraph_lazy_inclist_size()` returns the number of vertices in a lazy incidence list.
 - `igraph_bfs_simple()` now provides a simpler interface to the breadth-first search functionality.

### Changed

 - igraph now uses a CMake-based build sysyem.
 - GMP support can no longer be disabled. When GMP is not present on the system, igraph will use an embedded copy of Mini-GMP (PR #1549).
 - Bliss has been updated to version 0.75. Bliss functions are now interruptible. Thanks to Tommi Junttila for making this possible!
 - Adjacency and incidence lists:
   * `igraph_adjlist_init()` and `igraph_lazy_adjlist_init()` now require the caller to specify what to do with loop and multiple edges.
   * `igraph_inclist_init()` and `igraph_lazy_inclist_init()` now require the caller to specify what to do with loop edges.
   * Adjacency and incidence lists now use `igraph_vector_int_t` consistently.
 - Community detection:
   * `igraph_community_multilevel()`: added resolution parameter.
   * `igraph_community_fluid_communities()`: graphs with no vertices or with one vertex only are now supported; they return a trivial partition.
 - Modularity:
   * `igraph_modularity()` and `igraph_modularity_matrix()`: added resolution parameter.
   * `igraph_modularity()` and `igraph_modularity_matrix()` now support the directed version of modularity.
   * `igraph_modularity()` returns NaN for graphs with no edges to indicate that the modularity is not well-defined for such graphs.
 - Centralities:
   * `cutoff=0` is no longer interpreted as infinity (i.e. no cutoff) in `betweenness`, `edge_betweenness` and `closeness`. If no cutoff is desired, use a negative value such as `cutoff=-1`.
   * The `nobigint` argument has been removed from `igraph_betweenness()`, `igraph_betweenness_estimate()` and `igraph_centralization_betweenness()`, as it is not longer needed. The current implementation is more accurate than the old one using big integers.
   * `igraph_closeness()` now considers only reachable vertices during the calculation (i.e. the closeness is calculated per-component in the undirected case) (PR #1630).
   * `igraph_closeness()` gained two additional output parameters, `reachable_count` and `all_reachable`, returning the number of reached vertices from each vertex, as well as whether all vertices were reachable. This allows for computing various generalizations of closeness for disconnected graphs (PR #1630).
   * `igraph_pagerank()`, `igraph_personalized_pagerank()` and `igraph_personalized_pagerank_vs()` no longer support the `IGRAPH_PAGERANK_ALGO_POWER` method. Their `options` argument now has type `igraph_arpack_options_t *` instead of `void *`.
 - Shortest paths (PR #1344):
   * `igraph_average_path_length()` now returns the number of disconnected vertex pairs in the new `unconn_pairs` output argument.
   * `igraph_diameter()` now return the result as an `igraph_real_t` instead of an `igraph_integer_t`.
   * `igraph_average_path_length()`  and `igraph_diameter()` now return `IGRAPH_INFINITY` when `unconn=FALSE` and the graph is not connected. Previously they returned the number of vertices.
 - Trait-based random graph generators:
   * `igraph_callaway_traits_game()` and `igraph_establishment_game()` now have an optional output argument to retrieve the generated vertex types.
   * `igraph_callaway_traits_game()` and `igraph_establishment_game()` now allow omitting the type distribution vector, in which case they assume a uniform distribution.
   * `igraph_asymmetric_preference_game()` now accept a different number of in-types and out-types.
 - `igraph_subisomorphic_lad()` now supports graphs with self-loops.
 - `igraph_is_chordal()` and `igraph_maximum_cardinality_search()` now support non-simple graphs and directed graphs.
 - `igraph_realize_degree_sequence()` has an additional argument controlling whether multi-edges or self-loops are allowed.
 - `igraph_is_connected()` now returns false for the null graph; see https://github.com/igraph/igraph/issues/1538 for the reasoning behind this decision.
 - `igraph_lapack_ddot()` is renamed to `igraph_blas_ddot()`.
 - `igraph_to_directed()`: added RANDOM and ACYCLIC modes (PR #1511).
 - `igraph_topological_sorting()` now issues an error if the input graph is not acyclic. Previously it issued a warning.
 - `igraph_vector_(which_)(min|max|minmax)()` now handles NaN elements.
 - `igraph_i_set_attribute_table()` is renamed to `igraph_set_attribute_table()`.
 - `igraph_i_sparsemat_view()` is renamed to `igraph_sparsemat_view()`.

### Deprecated

 - `igraph_is_degree_sequence()` and `igraph_is_graphical_degree_sequence()` are deprecated in favour of the newly added `igraph_is_graphical()`.
 - `igraph_closeness_estimate()` is deprecated in favour of the newly added `igraph_closeness_cutoff()`.
 - `igraph_betweenness_estimate()` and `igraph_edge_betweenness_estimate()` are deprecated in favour of the newly added `igraph_betweenness_cutoff()` and `igraph_edge_betweenness_cutoff()`.
 - `igraph_adjlist_remove_duplicate()` and `igraph_inclist_remove_duplicate()` are now deprecated in favour of the new constructor arguments in `igraph_adjlist_init()` and `igraph_inclist_init()`.

### Removed

 - The following functions, all deprecated in igraph 0.6, have been removed (PR #1562):
   * `igraph_adjedgelist_init()`, `igraph_adjedgelist_destroy()`, `igraph_adjedgelist_get()`, `igraph_adjedgelist_print()`, `igraph_adjedgelist_remove_duplicate()`.
   * `igraph_lazy_adjedgelist_init()`, `igraph_lazy_adjedgelist_destroy()`, `igraph_lazy_adjedgelist_get()`, `igraph_lazy_adjedgelist_get_real()`.
   * `igraph_adjacent()`.
   * `igraph_es_adj()`.
   * `igraph_subgraph()`.
 - `igraph_pagerank_old()`, deprecated in 0.7, has been removed.
 - `igraph_vector_bool` and `igraph_matrix_bool` functions that relied on inequality-comparing `igraph_bool_t` values are removed.

### Fixed

 - Betweenness calculations are no longer at risk from integer overflow.
 - The actual cutoff distance used in closeness calculation was one smaller than the `cutoff` parameter. This is corrected (PR #1630).
 - `igraph_layout_gem()` was not interruptible; now it is.
 - `igraph_barabasi_aging_game()` now checks its parameters more carefully.
 - `igraph_callaway_traits_game()` and `igraph_establishment_game()` now check their parameters.
 - `igraph_lastcit_game()` checks its parameters more carefully, and no longer crashes with zero vertices (PR #1625).
 - `igraph_cited_type_game()` incorrectly rounded the attractivity vector entries to integers.
 - `igraph_residual_graph()` now returns the correct _residual_ capacities; previously it wrongly returned the original capacities (PR #1598).
 - `igraph_psumtree_update()` now checks for negative values and NaN.
 - `igraph_communities_spinglass()`: fixed several memory leaks in the `IGRAPH_SPINCOMM_IMP_NEG` implementation.
 - `igraph_incident()` now returns edges in the same order as `igraph_neighbors()`.
 - `igraph_modularity_matrix()` returned incorrect results for weighted graphs. This is now fixed. (PR #1649, thanks to Daniel Noom!)
 - `igraph_lapack_dgetrf()` would crash when passing `NULL` for its `ipiv` argument (thanks for the fix to Daniel Noom).
 - Some `igraph_matrix` functions would fail to report errors on out-of-memory conditions.
 - `igraph_maxdegree()` now returns 0 for the null graph or empty vector set. Previously, it did not handle this case.
 - `igraph_vector_bool_all_e()` now considers all nonzero (i.e. "true") values to be the same.
 - PageRank (PR #1640):
   * `igraph_(personalized_)pagerank(_vs)()` now check their parameters more carefully.
   * `igraph_personalized_pagerank()` no longer modifies its `reset` parameter.
   * `igraph_(personalized_)pagerank(_vs)`: the `IGRAPH_PAGERANK_ALGO_ARPACK` method now handles self-loops correctly.
   * `igraph_personalized_pagerank(_vs)()`: the result retuned for edgeless or all-zero-weight graphs with the `IGRAPH_PAGERANK_ALGO_ARPACK` ignored the personalization vector. This is now corrected.
   * `igraph_personalized_pagerank(_vs)()` with a non-uniform personalization vector, a disconnected graph and the `IGRAPH_PAGERANK_ALGO_PRPACK` method would return results that were inconsistent with `IGRAPH_PAGERANK_ALGO_ARPACK`. This happened because PRPACK always used a uniform reset distribution when the random walk got stuck in a sink vertex. Now it uses the user-specified reset distribution for this case as well.
 - Fixed crashes in several functions when passing a weighted graph with zero edges (due to `vector_min` being called on the zero-length weight vector).
 - Fixed problems in several functions when passing in a graph with zero vertices.
 - Weighted betweenness, closeness, PageRank, shortest path calculations and random walk functions now check if any weights are NaN.
 - Many functions now reject input arguments containing NaN values.
 - Compatibility with the PGI compiler.

### Other

 - Documentation improvements.
 - Improved error and warning messages.
 - More robust error handling.
 - General code cleanup to reduce the number of compiler warnings.
 - igraph's source files have been re-organized for better maintainability.
 - Debugging aid: When igraph is build with AddressSanitizer, the default error handler prints a stack trace before exiting.
 - igraph can now be built with an external CXSparse library.
 - The references to igraph source files in error and warning messages are now always relative to igraph's base directory.
 - When igraph is built as a shared library, only public symbols are exported even on Linux and macOS.

### Acknowledgments

 - Thanks to Daniel Noom for significantly expanding igraph's test coverage and exposing several issues in the process!

## [0.8.5] - 2020-12-07

### Changed

 - `igraph_write_graph_pajek()`: the function now always uses the platform-native line endings (CRLF on Windows, LF on Unix and macOS). Earlier versions tried to enforce Windows line endings, but this was error-prone, and since all recent versions of Pajek support both line endings, enforcing Windows line endings is not necessary any more.

### Fixed

 - Fixed several compilation issues with MINGW32/64 (PR #1554)
 - `igraph_layout_davidson_harel()` was not interruptible; now it is.
 - Added a missing memory cleanup call in `igraph_i_cattribute_combine_vertices()`.
 - Fixed a few memory leaks in test cases.

## [0.8.4] - 2020-11-24

### Fixed

 - `igraph_i_cattribute_combine_vertices()`: fixed invalid cleanup code that eventually filled up the "finally" stack when combining vertices with attributes extensively.
 - `igraph_hrg_sample()`: fixed incorrect function prototype
 - `igraph_is_posinf()` and `igraph_is_neginf()`: fixed incorrect result on platforms where the sign of the result of `isinf()` is not indicative of the sign of the input.
 - Fixed building with vendored LAPACK and external BLAS
 - Fixed building with XCode 12.2 on macOS

### Other

 - Documentation improvements
 - General code cleanup to reduce the number of compiler warnings

## [0.8.3] - 2020-10-02

### Added

 - `igraph_vector_binsearch_slice()` performs binary search on a sorted slice of a vector.

### Changed

 - `igraph_eigenvector_centrality()` assumes the adjacency matrix of undirected graphs to have twice the number of self-loops for each vertex on the diagonal. This makes the results consistent between an undirected graph and its directed equivalent when each edge is replaced by a mutual edge pair.

### Fixed

 - `igraph_isomorphic()` now verifies that the input graphs have no multi-edges (PR #1464).
 - `igraph_difference()` was creating superfluous self loops (#597).
 - `igraph_count_multiple()` was giving incorrect results for self-loops in directed graph (PR #1399).
 - `igraph_betweenness_estimate()`: fixed incorrect results with finite cutoff (PR #1392).
 - `igraph_count_multiple()` was giving incorrect results for self-loops in directed graph (PR #1399).
 - `igraph_eigen_matrix_symmetric()`: fixed incorrect matrix multiplication (PR #1379).
 - Corrected several issues that could arise during an error condition (PRs #1405, #1406, #1438).
 - `igraph_realize_degree_sequence()` did not correctly detect some non-graphical inputs.
 - `igraph_is_graphical_degree_sequence()`: fixed incorrect results in undirected case (PR #1441).
 - `igraph_community_leiden()`: fixed incorrect result when self-loops are present (PR #1476).
 - `igraph_eigenvector_centrality()`: fixed incorrect value for isolated vertices in weighted graphs.
 - `igraph_eigenvector_centrality()`: corrected the handling of self-loops.
 - `igraph_layout_reingold_tilford()`: fixed an issue where branches of the tree would sometimes overlap.

### Other

 - `igraph_degree_sequence_game()`: improved performance with `IGRAPH_DEGSEQ_SIMPLE_NO_MULTIPLE_UNIFORM` method.
 - Improved the robustness of the test suite.
 - Documentation improvements.
 - Improved error and warning messages.
 - Improved compatibility with recent versions of Microsoft Visual C.

## [0.8.2] - 2020-04-28

### Changed

 - Improved argument checking: `igraph_all_st_mincuts()` and `igraph_sir()`
 - Improved interruptibility: `igraph_sir()`

### Fixed

 - `igraph_community_leiden()`: fixed crash when interrupting
 - The tests are now more robust. Some incorrect test failures were fixed when
   running on i386 architecture, or when using different versions of external
   dependencies.

### Other

 - Improved error messages from `igraph_sir()`.
 - Improved compatibility with more recent versions of Microsoft Visual C.

## [0.8.1] - 2020-03-13

### Changed

 - Improved interruptability: `igraph_degree_sequence_game()`
 - Improved argument checking: `igraph_forest_fire_game()`
 - Updated the plfit library to version 0.8.1

### Fixed

 - `igraph_community_edge_betweenness()`: fix for graphs with no edges (PR #1312)
 - `igraph_bridges()` now handles multigraphs correctly (PR #1335)
 - `igraph_avg_nearest_neighbor_degree()`: fix for memory leak in weighted case (PR #1339)
 - `igraph_community_leiden()`: fix crash bug (PR #1357)

### Other

 - Included `ACKOWLEDGEMENTS.md`
 - Documentation improvements

## [0.8.0] - 2020-01-29

### Added

 * Trees

   - `igraph_to_prufer()` and `igraph_from_prufer()` convert labelled trees to/from Prüfer sequences
   - `igraph_tree_game()` samples uniformly from the set of labelled trees
   - `igraph_is_tree()` checks if a graph is a tree
   - `igraph_random_spanning_tree()` picks a spanning tree of a graph uniformly at random
   - `igraph_random_edge_walk()` returns the indices of edges traversed by a random walk; useful for multigraphs

 * Community detection

   - `igraph_community_fluid_communities()` detects communities based on interacting fluids
   - `igraph_community_leiden()` detects communities with the Leiden method

 * Cliques

   - `igraph_maximal_cliques_hist()` counts maximal cliques of each size
   - `igraph_maximal_cliques_callback()` calls a function for each maximal clique
   - `igraph_clique_size_hist()` counts cliques of each size
   - `igraph_cliques_callback()` calls a function for each clique
   - `igraph_weighted_cliques()` finds weighted cliques in graphs with integer vertex weights
   - `igraph_weighted_clique_number()` computes the weighted clique number
   - `igraph_largest_weighted_cliques()` finds the largest weighted cliques

 * Graph generators

   - `igraph_hsbm_game()` for a hierarchical stochastic block model
   - `igraph_hsbm_list_game()` for a more general hierarchical stochastic block model
   - `igraph_correlated_game()` generates pairs of correlated random graphs by perturbing existing adjacency matrix
   - `igraph_correlated_pair_game()` generates pairs of correlated random graphs
   - `igraph_tree_game()` samples uniformly from the set of labelled trees
   - `igraph_dot_product_game()` generates a random dot product graph
   - `igraph_realize_degree_sequence()` creates a single graph with a given degree sequence (Havel-Hakimi algorithm)

 * Graph embeddings

   - `igraph_adjacency_spectral_embedding()` and `igraph_laplacian_spectral_embedding()` provide graph embedddings
   - `igraph_dim_select()` provides dimensionality selection for singular values using profile likelihood

 * Isomorphism

   - `igraph_automorphism_group()` computes the generators of the automorphism group of a simple graph
   - `igraph_simplify_and_colorize()` encodes edge and self-loop multiplicities into edge and vertex colors; use in conjunction with VF2 to test isomorphism of non-simple graphs

 * Other

   - `igraph_bridges()` finds edges whose removal would disconnect a graph
   - `igraph_vertex_coloring_greedy()` computes a vertex coloring using a greedy algorithm
   - `igraph_rewire_directed_edges()` randomly rewires only the starting points or only the endpoints of directed edges
   - Various `igraph_local_scan_*` functions provide local counts and statistics of neighborhoods
   - `igraph_sample_sphere_surface()` samples points uniformly from the surface of a sphere
   - `igraph_sample_sphere_volume()` samples points uniformly from the volume of a sphere
   - `igraph_sample_dirichlet()` samples points from a Dirichlet distribution
   - `igraph_malloc()`, to be paired with the existing `igraph_free()`

### Changed

 - `igraph_degree_sequence_game()`: new method added for uniform sampling: `IGRAPH_DEGSEQ_SIMPLE_NO_MULTIPLE_UNIFORM`
 - `igraph_modularity_matrix()`: removed `membership` argument (PR #1194)
 - `igraph_avg_nearest_neighbor_degree()`: added `mode` and `neighbor_degree_mode` arguments (PR #1214).
 - `igraph_get_all_simple_paths()`: added `cutoff` argument (PR #1232).
 - `igraph_unfold_tree()`: no longer preserves edge ordering of original graph
 - `igraph_decompose()`: support strongly connected components
 - `igraph_isomorphic_bliss()`, `igraph_canonical_permutation()`, `igraph_automorphisms()`: added additional arguments to support vertex colored graphs (PR #873)
 - `igraph_extended_chordal_ring`: added argument to support direction (PR #1096), and fixed issue #1093.

### Other

 - The [Bliss isomorphism library](http://www.tcs.hut.fi/Software/bliss/) was updated to version 0.73. This version adds support for vertex colored and directed graphs.
 - igraph now uses the high-performance [Cliquer library](https://users.aalto.fi/~pat/cliquer.html) to find (non-maximal) cliques
 - Provide proper support for Windows, using `__declspec(dllexport)` and `__declspec(dllimport)` for `DLL`s and static usage by using `#define IGRAPH_STATIC 1`.
 - Provided integer versions of `dqueue` and `stack` data types.

[master]: https://github.com/igraph/igraph/compare/0.10.11..master
[0.10.11]: https://github.com/igraph/igraph/compare/0.10.10..0.10.11
[0.10.10]: https://github.com/igraph/igraph/compare/0.10.9..0.10.10
[0.10.9]: https://github.com/igraph/igraph/compare/0.10.8..0.10.9
[0.10.8]: https://github.com/igraph/igraph/compare/0.10.7..0.10.8
[0.10.7]: https://github.com/igraph/igraph/compare/0.10.6..0.10.7
[0.10.6]: https://github.com/igraph/igraph/compare/0.10.5..0.10.6
[0.10.5]: https://github.com/igraph/igraph/compare/0.10.4..0.10.5
[0.10.4]: https://github.com/igraph/igraph/compare/0.10.3..0.10.4
[0.10.3]: https://github.com/igraph/igraph/compare/0.10.2..0.10.3
[0.10.2]: https://github.com/igraph/igraph/compare/0.10.1..0.10.2
[0.10.1]: https://github.com/igraph/igraph/compare/0.10.0..0.10.1
[0.10.0]: https://github.com/igraph/igraph/compare/0.9.10..0.10.0
[0.9.10]: https://github.com/igraph/igraph/compare/0.9.9...0.9.10
[0.9.9]: https://github.com/igraph/igraph/compare/0.9.8...0.9.9
[0.9.8]: https://github.com/igraph/igraph/compare/0.9.7...0.9.8
[0.9.7]: https://github.com/igraph/igraph/compare/0.9.6...0.9.7
[0.9.6]: https://github.com/igraph/igraph/compare/0.9.5...0.9.6
[0.9.5]: https://github.com/igraph/igraph/compare/0.9.4...0.9.5
[0.9.4]: https://github.com/igraph/igraph/compare/0.9.3...0.9.4
[0.9.3]: https://github.com/igraph/igraph/compare/0.9.2...0.9.3
[0.9.2]: https://github.com/igraph/igraph/compare/0.9.1...0.9.2
[0.9.1]: https://github.com/igraph/igraph/compare/0.9.0...0.9.1
[0.9.0]: https://github.com/igraph/igraph/compare/0.8.5...0.9.0
[0.8.5]: https://github.com/igraph/igraph/compare/0.8.4...0.8.5
[0.8.4]: https://github.com/igraph/igraph/compare/0.8.3...0.8.4
[0.8.3]: https://github.com/igraph/igraph/compare/0.8.2...0.8.3
[0.8.2]: https://github.com/igraph/igraph/compare/0.8.1...0.8.2
[0.8.1]: https://github.com/igraph/igraph/compare/0.8.0...0.8.1
[0.8.0]: https://github.com/igraph/igraph/releases/tag/0.8.0