File python-rasterio.changes of Package python-rasterio

-------------------------------------------------------------------
Sun Jul 26 03:40:42 UTC 2020 - andy great <andythe_great@pm.me>

- Update to version 1.1.5.
* Earlier versions of rasterio set the CHECK_WITH_INVERT_PROJ 
  config option to `True` by default. This is very rarely necessary
  and broke the GRIB format driver (#1248), so we no longer set 
  this option (#1942). Users of rasterio 1.1.5 in combination with 
  GDAL 1.11 and PROJ 4.8 may see some small differences, compared 
  to rasterio versions < 1.1.5, in results of warping global 
  datasets.
* WarpedVRT can properly handle two use cases that weren't ruled 
  out in version 1.1.4: simple scaling of datasets and control 
  over the scaling of reprojected output (#1921, #1936).
* The error in making boundless reads of datasets opened using the
  OVERVIEW_LEVEL reported in #1929 has been resolved by #1939.
* The pixel shift in reads from datasets reported in the user 
  discussion group and #1932, has been fixed (#1938).
* We have extended the signature of merge's method function 
  (#1933).
* The MemoryFile implementation has been improved so that it can
  support multi-part S3 downloads (#1926).
* Members of the Resampling enum with a value > 7 can only be used
  in warp operations (#1930). We now raise a 
  ResamplingAlgorithmError if they are used with non-warp read and
  writes.
* To help users of poetry, the conditional requirements in setup.py 
  have been changed to use PEP 496 environment markers exclusively 
  (#1777).

-------------------------------------------------------------------
Mon Nov  4 17:34:23 UTC 2019 - Bruno Friedmann <bruno@ioda-net.ch>

- Update to version 1.1.0 see changes for all changes
  + Support for Python 2 is deprecated and a warning is raised from the
    rasterio.compat module. 
  + This module will be deleted in a future version.
- Fix build by adding proj (proj.db) as buildrequires

-------------------------------------------------------------------
Wed Mar  6 19:23:58 UTC 2019 - Todd R <toddrme2178@gmail.com>

- Update to version 1.0.21
  * Fix for bug in implementation of the pickle protocol.
- Update to version 1.0.20
  * Fix for an unchecked NULL pointer introduced in 1.0.19 that could result in
    segmentation fault on import of rasterio._env on Linux when GDAL data files
    are not available (for example, with wheels as on PyPI).
- Update to version 1.0.19
  * Do not set GDAL_DATA and PROJ_LIB when data files can be found at their built
    in locations.
  * Add linear_units property to CRS.
  * Ensure that AWS_NO_SIGN_REQUESTS is sufficient for accessing public S3
    datasets and that import of boto3 is not required.
  * An out_dtype parameter has been added to DatasetReaderBase.read, enabling
    on-the-fly casting of raster values to the desired type. This uses
    the same latent feature used to get support for "int8" data.
  * Restore pickle protocol for CRS, using WKT as state.
  * Support for signed 8-bit integer datasets ("int8" dtype) has been added.
- Update to version 1.0.18
  * Fix a regression.
- Update to version 1.0.17
  * Fix a regression in evaluation of CRS equality.
- Update to version 1.0.16
  * A bug preventing GCPs from being created with new BufferedDatasetWriter
    instances has been fixed.
  * A previously unreported bug preventing BufferedDatasetWriters from being
    opened in r+ mode has been fixed.
  * A regression in creating CRS objects from PROJ4 strings that include
    "+wktext" has been fixed.
  * Regressions in str representations of empty CRS objects and the handling of
    unreferenced datasets in rasterio._base have been fixed.
  * GDAL seems to work best if GDAL_DATA is set as early as possible. Ideally it
    is set when building the library or in the environment before importing
    Rasterio, but for wheels we patch GDAL_DATA into os.environ when rasterio.env
    is imported.
- Update to version 1.0.15
  * Google cloud storage support was *not* in fact added in 1.0.14, but is
    present in 1.0.15.
- Update to version 1.0.14
  * The _CRS class has been refactored so that a WKT representation, rather than
    PROJ4 representation, is the canonical form.
  * Native support for Google Cloud Storage using "gs://" URLs has been added.
  * On entering a dataset context (DatasetBase.__enter__) a new anonymous GDAL
    environment is created if needed and is entered. This makes `with
    rasterio.open(...) as dataset:` roughly equivalent to `with
    rasterio.open(...) as dataset, Env():`. This helps prevent bugs when datasets
    are created and then used later or are used in different scopes.
- Update to version 1.0.13
  * Fix a buffer dttype mismatch on Windows introduced in 1.0.12.
- Update to version 1.0.12
  * Rasterio's lower level I/O functions now take Py_ssize_t index arrays and
    will raise exceptions for type mismatches instead of swallowing the
    exceptions.
  * The _boundless_vrt_doc function's background layer no longer needs an
    in-memory dataset and a performance regression has been
    reversed. We've also found a performance improvement in masked boundless
    reads in the case that the source data is entirely valid.
  * The signature of the private _boundless_vrt_doc function in rasterio.vrt has
    changed. Its background keyword argument now takes an int or float, a new
    masked keyword argument has been added, and the function returns a unicode
    str instead of ascii-encoded bytes.
  * The copy and copyfiles functions of rasterio.shutil now raise an exception
    when the source and destination paths identify the same dataset.
- Update to version 1.0.11
  * Prevent needless warning when making boundless reads with a fill value.
  * The GDAL band cache is flushed at the top of build_overviews
  * Options --gdal-data and --proj-data have been added to the rio-env command so
    that users of Rasterio wheels can get paths to set GDAL_DATA and PROJ_LIB
    environment variables.
  * The attempt to make GDAL and PROJ support file discovery and configuration
    automatic within methods of the CRS class has been reverted. Users must
    execute such code inside a `with Env()` block or set the GDAL_DATA and
    PROJ_LIB environment variables needed by GDAL.
- Update to version 1.0.10
  * Avoid segmentation fault when OSRGetAuthority* functions return null
    pointers, making Rasterio more robust when reading CRS that don't map
    perfectly to PROJ.
  * Ensure that GDAL and PROJ support files can be found in Rasterio wheels when
    we call CRS methods.
  * Accomodate the Esri flavor of projection WKT.
  * Add missing raster offsets and scales properties.
- Update to version 1.0.9
  * Delegate test of the environment for existing session credentials to the
    session class to generalize credentialization of GDAL to cloud providers
    other than AWS. The env.hascreds function is no longer used in
    Rasterio and has been marked as deprecated.
  * Switch to use of botocore Credentials.get_frozen_credentials.
  * Numpy masked arrays with the normal Numpy mask sense (True == invalid) are
    now supported as input for feature.shapes(). The mask keyword argument of the
    function keeps to the GDAL sense of masks (nonzero == invalid) and the
    documentation has been improved to communicate this.
  * The defaults for WarpedVRT's `src_nodata` and `nodata` keyword arguments are
    no longer None, but are markers for the default behavior: inheritance of
    nodata values from the VRT's source dataset. Values of `None` for these
    keyword arguments means that the VRT does not inherit from the source and has
    no nodata value at all.
  * WEBP has been added to the Compression enum. Exploitation of this new
    compression mode will require GDAL 2.4.
  * Rasterio now check for two AWS environment variables before creating a
    session.
  * The ensure_env_with_credentials decorator which wraps rasterio.open will no
    longer clobber the credentials in a surrounding environment or seek
    credentials when they already exist.
  * The comparison of CRS objects to dicts and strs that we pledged to remove at
    1.0 has at last been removed.
- Update to version 1.0.8
  * Datasets contained in MemoryFile buffers are now opened in r+ or w+ mode
    instead of r or w.
  * The namedtuple _asdict method is overridden in BoundingBox to work around a
    bug in Python 3.4.3. This unblocks creation of manylinux1
    wheels for Python 3.4.
- Update to version 1.0.7
  * Use the non-resolving path form of files_inout_arg in rio-convert and
    rio-shapes.
  * Filling the empty regions of boundless reads was too slow in some cases and
    a faster solution has been found.
  * Require cligj>=0.5 for compatibility with click 7.0.
  * Precisely specify CLI option and argument names for click 6.x and 7.0
    compatibility.
- Update to version 1.0.6
  * If the build_overviews method of a dataset is passed a list of factors that
    specify more than one 1x1 pixel overview, Rasterio raises an
    exception.
  * Calling calculate_default_transform for large extents should no longer result
    in an out of memory error. The rio-warp command should
    also now run more quickly and with a smaller memory footprint.
  * We have a more general fix for the problem of filling the empty regions of
    boundless reads.
- Update to version 1.0.5
  * The fill value for boundless reads was ignored in Rasterio versions 1-1.0.4
    but now applies.
  * An invalid shortcut has been eliminated and Rasterio now produces a proper
    mask in the boundless masked read case.
  * Loss of a row or column in geometry_window() and mask() has been fixed.
- Update to version 1.0.4
  * Boundless reads of datasets without a coordinate reference system have been
    fixed.
  * A y-directional error in disjoint_bounds has been fixed.
  * Prevent geometries from being thrown near projection singularities.
  * Missing --aws-no-sign-requests and --aws-requester-pays options added to
    the main rio command.
  * Add missing bilinear, cubic spline, lanczos resampling modes for overviews.
  * Raise ValueError if get_writer_for_driver() is called without a driver
    name.
  * Windows are now frozen so that they are hashable.
  * Use of InMemoryRaster eliminates redundant code in the _warp module.
  * Generalize sessions to support cloud providers other than AWS.
- Update to version 1.0.3
  * A regression in GeoJSON source handling in rio-rasterize has been
    fixed.
  * Rasterization of linear rings is now supported.
  * The input file handler of rio-rasterize has been changed so that it does not
    reject GeoJSON files.
  * Linear rings have been added as acceptable input for rio-rasterize and other
    commands and functions.
- Update to version 1.0.2
  * The output of calculate_default_transform() can be fixed to output dimensions
    as well as to output resolution.
  * In using rio-warp, the --src-bounds option can now override the bounds of the
    source dataset when --dimensions is used.
  * Bounds of rotated rasters are now calculated correctly.
  * A band indexing bug in reproject() has been fixed.
- Update to version 1.0.1
  * Bounding envelopes are densified at higher precision in transform_bounds.
  * Add LERC compression to enums.Compression.
  * The reproject() function now passes dst_alpha properly to _reproject(), which
    unlocks materialization of warp destination alpha band masks.
  * The --dimensions and --src-bounds options of rio-warp can be used together
    as expected.
- Update to version 1.0.0
  * C source files are now generated using Cython 0.28.3, which gives us support
    for Python 3.7.
  * An integer overflow when multiplying the warp memory limit has been
    fixed.
  * CRS based EPSG codes can now report that code using a new `to_epsg()` method.
  * A regression in rasterio.warp.reproject has been reported, confirmed,
    and fixed.
  * The rasterio.vfs module has been removed.
  * Reading array data or masks from datasets opened in "w" mode is prohibited.
  * Property set_* methods of datasets have been removed. The appropriate
    properties should be used instead. For example, ``dataset.crs = "EPSG:4326"``
    instead of ``dataset.set_crs("EPSG:4326")``.
  * Rasterio ignores creation options saved in the metadata of datasets
    by Rasterio versions < 1.0b1.
  * Internal Env() in `rasterio.open` has been replaced with an environment
    ensuring decorator. The same decorator ensures that credentials are
    obtained when functions from `rasterio.shutils` are called.
  * Input file arguments for all CLI commands are now parsed and validated in
    a uniform manner.
  * Local loggers have all been changed to `getLogger(__name__)` in rasterio.rio
    module.
  * We now raise an exception when WarpedVRT is asked to add an alpha band and
    one already exists for the VRT.
  * The warp memory limit configuration available in gdalwarp has been added
    to `reproject()` and `WarpedVRT`.
  * Avoid boundless reads when sampling pixels outside a dataset's extent.
  * The logic behind rio-warp's --target-aligned-pixels has been factored into
    a new `aligned_target()` function in the warp module.
  * A regression in the VSI path of some zip:// URLs has been fixed.
  * Transform, coordinate reference system, and ground control points are
    properly set when opening a dataset in w+ mode.
  * `boundless=True` reads from warped VRTs are disallowed. Instead, VRTs should
    be constructed so that only data bounded by their extents is read. In
    exchange, boundless reads for non-VRT datasets are faster and more robust.
  * Comparison of CRS objects to str or dict is deprecated.
  * Added `out`, `out_shape`, `window`, `resampling` keyword parameters to
    the `dataset_mask` method.
  * CRS.from_string can now take WKT strings as input.
  * Tests have been added to confirm that reading from a WarpedVRT will use the
    source dataset's masks and overviews.
  * Expectations of warp tests that use resampling algorithms other than nearest
    neighbor have been corrected.
  * WarpedVRT.crs now returns a CRS, not a str.
  * A Windows-crashing bug in the _warp module has been fixed.
  * The signatures of all Dataset constructors have changed: these method now
    take instance of ParsedPath or UnparsedPath from rasterio.path and do not
    take strings.
  * The signatures of get_writer_for_path and get_writer_for_driver have 
    changed in the same way.
  * Whether to print GeoJSON feature sequences or a GeoJSON feature collection
    from rio-shapes is now controlled by a ``--sequence/--collection`` option.
    A sequence is now the default.
  * Comparing empty ``CRS`` objects returns ``True``.
  * Deprecated window functions have been removed from the rasterio module. Their
    replacements are in rasterio.windows.
  * The deprecated r- mode has been removed from ``rasterio.open()``.
  * A bytearray is no longer allowed as input to the ``MemoryFile`` constructor.
    Users must convert byte arrays to ``bytes before calling ``MemoryFile()``.
  * Signatures of private functions and classes in _features, _warp, _io have
    been changed to always take instances of Affine instead of GDAL geotransform
    arrays.
  * Calling `rasterio.Env()` no longer sets hidden default config options
    CHECK_WITH_INVERT_PROJ=True and GTIFF_IMPLICIT_JPEG_OVR=False.
  * Rasterio no longer saves creation options in metadata on created datasets.
  * In the rasterio.open function: we have removed 'affine' as an alias for
    'transform'.
  * In the _base.DatasetBase class, and thereby all dataset objects: we have
    removed the mask_flags and affine properties. Instead, users must use the
    mask_flag_enums and transform properties.
  * In the _io.DatasetReaderBase class, and thereby all dataset objects: we have
    removed the read_mask method. Users must use the read_masks method.
  * In the features.rasterize function: we have removed 'replace' and 'add' as
    aliases for ``MergeAlg.replace`` and ``MergeAlg.add``.
  * In the profiles.Profile class: we have removed 'affine' as alias for
    'transform' and have removed the `__call__` method. The class is no longer
    callable. Users must use the mapping protocol to copy and update profiles.
  * In the windows.WindowMethodsMixin class, and thereby all dataset objects: we
    have removed the 'boundless' keyword arg from the window and from_bounds
    methods.
  * In the windows.Window class: we have removed the num_cols and num_rows
    properties, the __getitem__ method, and the from_offlen and from_ranges class
    methods.
  * The rasterio.vfs module and its functions are being replaced by new functions
    in rasterio.path.
  * Reading array data or masks from datasets opened in "w" mode will be
    prohibited in Rasterio 1.0. We are warning about this now.
  * Property set_* methods of datasets will be deprecated in 1.0. The appropriate
    properties should be used instead. For example,
    ``dataset.crs = "EPSG:4326"`` instead of ``dataset.set_crs("EPSG:4326")``.
  * Rasterio 1.0 will ignore creation options saved in the metadata of datasets
    by Rasterio versions < 1.0b1. Users can opt in to this behavior now by
    setting RIO_IGNORE_CREATION_KWDS=TRUE in their environments.
  * New ParsedPath and UnparsedPath classes have been added to improve input
    file handling and validation. The existing parse_path and vsi_path functions
    have been rewritten to use these new classes and have been moved to the new
    rasterio.path module. The signature of rasterio.open has not been changed
    and users of Rasterio will be unaffected.
  * set_band_unit and set_band_description methods have been added to
    dataset writers/updaters to support rio-edit-info.
  * A tools module has been added to collect file-based functions and supporting
    classes.
  * ``rasterio.features.dataset_features()`` has been extracted from
    rio-shapes.
  * The block_shapes and files properties of datasets now return lists instead
    of tuples.
  * The ``--nodata`` option of rio-merge is now passed to the output file.
  * The precision keyword arguments for methods in the windows and transform
    modules now default to None, which means no rounding of values by
    default.
  * The pixel_precision keyword arguments for methods in the windows module now
    default to None.
  * Rasterio functions and methods, even the private ones, now always take
    instances of Affine as transform parameters.
  * Importing the features, fill, io, and warp modules would set the
    CHECK_WITH_INVERT_PROJ=True and GTIFF_IMPLICIT_JPEG_OVR=False config options
    in a hidden way.
  * Add Debian data directory to set of well-known paths to search.
  * Pass precision parameter from ``merge`` to ``from_bounds``.
  * Added the missing w+ mode.
  * Rotated datasets are now masked properly.
  * Geometry collections are flattened in ``rasterize`` to work around
    incidental holes created by GDAL.
  * Add missing support for Python geo protocol in features module.
  * ``mask()`` now takes the ``indexes`` keyword parameter that is common
    elsewhere in the package.
  * Sharing of dataset handles may be disabled for multithreaded applications
    using the new `sharing` keyword argument of `rasterio.open`.
  * ``WarpedVRT()`` properly sets a CRS on the internal VRT handle when
    that VRT is produced by ``GDALCreateVRT()``.
  * ``rasterio.open()`` now accepts URLs with GET parameters.
  * Specific drivers and options can be used in ``rasterio.open()``.
  * ``rasterize()`` now supports GDAL's 'add' algorithm.
  * GDAL version checking has been consolidated and improved.
  * Pad boundless read VRTs so we don't overflow due to rounding.
  * Boundless reads of WarpedVRTs are enabled by setting an effective crs.
  * Raise ValueError when invalid width/height are passed to the
    ``InMemoryRaster`` constructor.
  * GDAL errors that are turned into Rasterio exceptions are no longer also sent
    automatically to the logger. They may, of course, be logged by user code.
  * Raster width and height were swapped in ``geometry_window()``.
  * ``rasterio.copy()`` moved to ``rasterio.shutil.copy().
  * ``DatasetBase.colorinterp`` is now a setter/getter that operates on
    a sequence of ``ColorInterp.<enum>`` in band order.
  * Addition of ``rasterio.shutil.delete()`` and ``$ rio rm`` for deleting
    datasets.
  * Addition of ``rasterio.shutil.exists()`` for determining if a dataset
    exists.
  * Addition of ``rasterio.shutil.copyfiles()`` to copy files associated
    with a datset from one location to another.
  * Allow setting band color interpretation in a ``set_colorinterp()`` method
    and in ``$ rio edit-info --colorinterp``.
  * Reprojection of numpy arrays using GCPs has never worked due to faulty code,
    but has been fixed.
  * A circular import situation has been resolved by moving dataset mixin classes
    from ``rasterio.io`` to ``rasterio.windows`` and ``rasterio.transform``.
  * Prevent a segmentation fault in ``rasterio.features.rasterize()`` by guarding
    against unsupported GeoJSON input.
  * In the GeoJSON output of rio-blocks, the windows are now JSON
    representations of the ``Window`` class.
  * The ``rasterio.windows.Window`` class no longer derives from ``tuple``.
    Comparisons like ``Window(0, 0, 1, 1) == ((0, 1), (0, 1))`` are no
    longer possible. Instead, call the ``.toranges()`` method of the 
    former or coerce the latter using ``Window.from_ranges``.
  * The ``rasterio.windows.from_bounds`` function now always returns unbounded,
    meaning uncropped, Windows. For example, if given a bounding box entirely
    to the west and north of the geotransform's origin, the result will be a 
    Window entirely to the left and above the dataset's row and column origin.
    Passing a ``boundless`` keyword argument to this function will result in a 
    warning.
  * The ``TransformMethodsMixin.ul`` method, deprecated since 0.36, has been
    removed.
  * Float precision read/write windows are now supported throughout
    Rasterio.
  * Use of old style range tuples as windows is deprecated and warned
    against.
  * Addition of dataset ``block()`` method.
  * Added access to subdatasets, both in the API and CLI (see `rio info
    --subdatasets`).
  * We always reacquire the GIL when GDAL calls the rasterio logging error
    handler.
  * Initialization of an unneeded array, sometimes contributing to memory
    exhaustion, has been eliminated.
  * Rasterio's boundless reading has had a compositing bug. We are resolving
    it by using a VRT internally and relying upon the VRT's windowing and
    compositing logic.
  * AWS credentials set in an ``Env`` are no longer clobbered by credentials
    from a default boto3 session inside ``rasterio.open``.
  * NaN can now be used as a nodata value in ``merge`` and when setting nodata
    from `rio edit-info`.
  * Removed `**options` argument from `rasterio.warp.transform()`.      Transformer
    options should be set explicitly inside the currently active
    `rasterio.env.Env()`.      This argument was initially added in `1.0a1`.
  * The function `transform_geom()` now always splits geometries at the anti-
    meridian.
  * The `crs` property of a dataset is now `None` instead of `CRS()` when the
    dataset's coordinate reference system is undefined.
  * Enable setting and getting `GDAL_CACHEMAX`.
  * Integer config option values are now properly encoded and decoded.
  * A `from_wkt()` CRS constructor has been added.
  * Targeting aligned pixels in rio-warp.
  * A new `WarpedVRT` class that surfaces GDAL's warp-on-demand VRT features. No
    XML editing is required.
  * `dtypes.get_minimum_dtype()` now returns proper value for uint* types.
  * Rasterio now uses `OSRRelease()` to avoid destroying shared spatial 
    reference system objects.
  * GDAL_SKIP and GDAL_DRIVER_PATH options, when needed, are now set before
    driver registration.
  * A failure of reprojection to an array when source dataset bands have an
    index higher than 1 has been fixed.
  * Secrets kept in GDAL config options could have been leaked via the Python
    logger. AWS keys and tokens have always been redacted, but other options like
    GDAL_HTTP_USERPWD were not. Logging of GDAL config options has been removed.
  * Use of Rasterio with Python threads has not been well tested in
    previous versions and a bug involving shared GDAL environment
    state was found. Rasterio now uses thread local variables to track
    GDAL environment state and uses the appropriate GDAL API functions to
    isolate the environments of child threads from each other while
    permitting inheritance from the main Python thread.
    Tests using both `ThreadPoolExecutor` and `ProcessPoolExecutor` have
    been added to guard against regresion.
  * Use __stdcall for GDAL callbacks on Windows.
  * Tear down Env more carefully, taking particular care not to
    reinstate default options when a child environment is exited.
  * Always close open file handles when exiting context managers returned by
    `rasterio.open()` when file objects are given as arguments.
  * Properly append, not extend, shim module to extension list.
  * Order of input datasets is preserved by rio-calc.
  * `MemoryFile()` can now take a file object parameter or bytes. The file should
    be opened in binary mode and its contents are read immediately into bytes.
    This form of argument is for convenience only; the provided file is not 
    accessed in any other way.
  * A `ZipMemoryFile` class derived from `MemoryFile` has been added. It is
    created from a sequence of bytes representing a zip file. Its `open()` method
    can take a path, allowing access to files within the zip file.
  * For convenience of users, Rasterio's `open()` function now takes Python file
    objects as well as filenames and URLs. These file objects have their
    contents read into `MemoryFile` instances within the `open()` call, so this
    feature is equivalent to calling `MemoryFile(fp).open()`.
  * GDAL/OGR function definitions consolidated in gdal.pxi.
  * Shim APIs have been created to provide support for GDAL 2.0 and 2.1 features
    that may be absent in a system's GDAL installation.
  * Shim API C modules are included in source distributions.
  * New binary wheels using version 1.2.0 of sgillies/frs-wheel-builds.
  * Geographic bounds designated by a `--geographic` option may now be
    passed to `rio clip`.
  * The `rio edit-info` command's new `--unset-nodata` option can be used to
    delete a dataset's nodata value.
  * Passing a `--co nodata=null` option to CLI commands will create
    datasets without any nodata value set. The `--co` option name follows
    GDAL's command line utilities. A `--profile` alias for this option has
    also been added.
  * The `DatasetBase` class is now surfaced in `rasterio.io`. This is not too
    promote the practice of subclassing `DatasetBase` but to help users that
    would like to use type annotations or make `isinstance()` assertions.
  * Ground control points may be read from and written to datasets (when the
    format permits) and may be used when reprojecting arrays with `reproject()`
    or on the command line with rio-warp. A new rio-gcps command
    has been added, it can print a dataset's GCPs as GeoJSON Feature sequences
    or FeatureCollections.
  * A `BytesIO` like `MemoryFile` class has been added for an improved Python
    interface to GDAL's in-memory filesystem.
  * Error handlers are pushed to GDAL on entering a GDAL environment and popped
    when exiting.
  * The GIL is released when calling GDAL's ChunkAndWarp().
  * transform_geom() would fail when encountering geometries with z coordinate
    values, but now handles them properly.
  * show_hist() failed when nans were present in data and this has been
    fixed with the use of nanmin() and nanmax().
  * A new `mask_flag_enums` attribute has been added to dataset objects.
  * Support for complex data types has been enabled.
  * Reduce log level when trying to identify a CRS as an EPSG code.
  * Acquire GIL when executing the error handler.
  * Raster data I/O involving ndarray views has now been corrected.
  * Support for single band index in `sample()`, which was advertised in the
    docstring but not implemented, has been added.
  * The rio-overview CLI command no longer opens datasets in 'r+' mode when
    listing overviews (with the `--ls` option).
  * Implicit JPEG overviews for JPEG-compressed GeoTIFF files have been disabled
    in Rasterio's default environment, which means decimation is slower but
    shows fewer JPEG compression artifacts.
  * Data type-specific functions such as `io_uint8()` have been eliminated from
    the private `_io` module and replaced by generic `io_band()`,
    `io_multi_band()`, and `io_multi_mask()` functions.
  * New introduction, guidance on switching from GDAL's Python bindings and much
    more.
  * The sense of rio-warp's `--bounds` options has been changed from source
    bounds to destination bounds.
  * Extracted Rasterio's public dataset classes into `rasterio.io`.
  * All Rasterio functions with will raise an exception if a GDAL style
    geotransform sequence is passed as `transform`.
  * The `affine` attribute of dataset objects is deprecated.
  * The `affine` member of dataset `profile` has been removed.
  * The `rasterio.drivers()` context manager has been removed. It is replaced by
    `rasterio.Env()`.
  * The `read_band()` method of dataset objects has been removed. It is replaced
    by `read()`.
  * The `read_mask()` method of dataset objects has been removed. It is replaced
    by `read_masks()`.
  * `get_data_window()`, `window_union()`, `window_intersection()`, and
    `windows_intersect()` have been removed from the `rasterio` module. They
    have been moved to `rasterio.windows`.
  * The `rasterio.tool` module has been removed. Its functions have been 
    relocated to `rasterio.plot` and `rasterio.rio.insp`.
  * The `rasterio.tools` module has been removed. Its functions have been
    relocated to `rasterio.mask` and `rasterio.merge`.
  * The `rasterio.warp.RESAMPLING` enum has been replaced by
    `rasterio.warp.Resampling`.
  * The signatures of `rasterio.features.sieve()` and
    `rasterio.features.rasterize()` have been changed: the `output` argument
    is deprecated and is replaced by `out`.
  * In the CLI, the `-o` option no longer causes existing output files to be
    overwritten.
  * Added `descriptions` and `units` attributes to dataset objects, exposing
    the descriptions and units of dataset bands.
  * Creation of a `Window` class to abstract Rasterio's window selection
    tuples.
  * Performance of `reproject()` has been increased by using an approximate
    transformer as with gdal_translate.
  * `Band` objects may now represent one or more dataset bands and multiband 
    reprojection of imagery kept in a GDAL dataset is now possible.
  * `rasterize()` now checks that the `out_shape` is 2D before continuing.
  * Named AWS credential profiles are now properly set up up in the main CLI
    command's environment.
  * In the CLI, a GDAL environment is now configured by the top level command
    and then passed in a Click context to subcommands.
  * Removal of redundant copies of GDAL geotransform detection code.
  * Window methods of dataset objects have been extracted into functions of
    a new `rasterio.windows` module.
  * `warp.project()` and rio-warp may lose one or more columns of data when 
    resampling to lower resolution.
  * rio-merge now copies colormaps to output files.
  * The correct `--force-overwrite` and `--output` usage is now provided in the
    case of a `FileOverwriteError`.
  * Passing undefined CRS to `reproject` no longer causes a segfault.
  * GDAL's invert projection check is always used by default in reprojecting.
  * Forward slashes are always used for GDAL VSI paths (`/vsizip/` etc) instead
    of `os.path.sep`.
  * Contributing guidelines have been added.
  * Axis order has been corrected in image processing doc.
  * A framework for comprehensive documentaton has been created.
  * `--src-nodata` and `--dst-nodata` options for rio-warp.
  * `read()` and `read_masks()` take an `out_shape` argument for decimated reads.
  * Color interpretation of bands added to rio-info output.
  * Dataset objects have a new per-dataset mask property: `dataset_mask`.
  * Utility functions for rehaping and plotting arrays have been added to
    `rasterio.plot`.
  * New `CRS` class like our old crs dicts, but with methods attached.
  * setup.py has new install extras: '[plot]', and '[all]'.
  * We've standardized on `import numpy as np` throughout the
    project.
  * The `five` module has been renamed to `compat`.
  * More coverage, more xfailing tests to mark known bugs.
  * Bug fix: restore support for URI-like GDAL dataset names such as
    'NETCDF:foo:bar'.
  * Bug fix: ensure GDAL environment is initialized for `transform_bounds()` as
    well as the other functions in `rasterio.warp`. In implementation, we
    have done this with a function decorator.
  * Bug fix: added rasterfill.cpp to MANIFEST.in so it is included in source
    distributions no matter the build system's GDAL version.
  * Requirements: affine requirement upped to >=1.3.0.
  * Bug fix: passing an empty JSON object to `crs.from_string()` raises CRSError
    instead of passing silently.
  * Bug fix: GDAL errors are no longer written to stderr; we no longer undefine
    error handlers.
  * Bug fix: the Rasterio library only configures a NullHandler, applications
    must configure their own handlers to see Rasterio's log messages.
  * Bug fix: AWS credentials are only sought by Rasterio when s3:// URLs are
    passed to `rasterio.open()`.
  * Bug fix: window comparison functions now raise a ValueError when windows do
    not intersect instead of returning an empty sequence.
  * Bug fix: upgrade from deprecated Numpy usage in `read()` by explicitly
    converting window offsets to ints.
  * Refactoring: window comparison functions may now take a variable number of
    windows as positional arguments in addition to a sequence of windows.
  * Refactoring: logging is much finer grained now because we've changed to the
    `logger = logging.getLogger(__name__)` pattern throughout Rasterio.
  * Refactoring: replaced old `drivers()` implementation with a new `Env` class
    and more consistent usage of it through the library and command line
    interface.
-  * Bug fix: S3 support was found missing in several of the CLI commands 
    mentioned below. This is corrected in 0.34 and we have the tests to prove it.
  * Bug fix: YCbCr JPEG-in-TIFF files no longer break rio-info.
  * New feature: the ability to read dataset metadata and imagery from S3 objects
    is an extra feature that can be installed like this `pip install -U
    rasterio[s3]`. AWS Credentials are handled by boto3 (and botocore) and so can
    be provided by environment variables, session arguments, `~/.aws/credentials`
    file, or EC2 instance metadata. S3 access is enabled in the following CLI
    commands: rio-clip, rio-info, rio-insp, rio-bounds, rio-shapes, rio-sample
    and may be expanded in future versions. S3 objects are identified on the
    command line and in API functions by URIs following the pattern
    `s3://bucket/object`. Extra thanks to Rob Emanuele and Even Rouault for
    helping on this one.
  * New feature: new and improved documentation coming soon to a website near
    you.
  * Refactoring: commands for the rio CLI have been moved to their own
    modules so that they're easier to find.
  * Refactoring: we've changed our primary pattern for checking errors set by
    GDAL API functions.
- Update to version 0.32.0.post1
  * No changes to the library in this post-release version, but there is a
    significant change to the distributions on PyPI: to help make Rasterio more
    compatible with Shapely on OS X, the GDAL shared library included in the
    macosx (only) binary wheels now statically links the GEOS library.
- Update to version 0.32.0
  * Bug fix: geometry factories and warp operations are properly deallocated
    in normal and error situations.
  * Bug fix: a code block in rio-merge's help has been better formatted.
  * Bug fix: the rasterio.vfs module is imported in __init__.py to assist
    cx_Freeze.
  * Bug fix: old usage of `read_band()` has been replaced by `read()` throughout
    the docs.
  * Bug fix: accidental overwriting of existing files is now prevented by the
    `resolve_inout()` function in `rasterio.rio.helpers`. Commands that take
    one or more input files plus an output file should use this helper and force
    overwrite either by using a `--force-overwrite` option or by using the
    `-o output` option, which implicitly forces overwriting.
  * Bug fix: missing support for NaN nodata value in rio-warp added.
  * Bug fix: missing documentation of `rasterize()`'s `fill` parameter added.
  * Bug fix: raster dataset bounds are densified before transforming so that
    the projected output of rio-bounds is correct.
  * Bug fix: add 'line' to the `Interleaving` enum.
  * Bug fix: convert `matplotlib` import errors to a `RuntimeWarning`.
  * Bug fix: deallocate CPL strings in error cases.
  * Bug fix: non-invertable affine transforms are prevented using
    `__future__.division`.
  * Bug fix: rio-warp clips output regions to the limits of the destination
    CRS unless disabled with `--no-check-invert-proj`.
  * New feature: the functionality previously available only in rio-mask is now
    available as `rasterio.tools.mask.mask()`.
  * New feature: raster bounds are used to label axes in `rasterio.tool.show()`.
  * New feature: GDAL's suggested warp bounds algorithm is wrapped and exposed
    for use in `warp()` and rio-warp.
  * Breaking change: align rio-warp's `--bounds` option with rio-merge's: these
    are in destination CRS units.
- Update to version 0.31.0
  * Warn when rasters have no georeferencing and when the default identity
    transform will be applied by GDAL.
  * Build OS X wheels using numpy>=1.10.2.
  * When reading image windows in previous versions, given a window with
    ((row_start, row_stop), (col_start, col_stop)) if the stop index is greater
    than the width/height the start index effectively shifts as well. This can
    manifest itself in pixel misalignment if, e.g. you read block windows with
    a bit of padding to avoid edge effects. Now the window offsets are determined
    solely by row_start and col_start..
- Update to version 0.30.0
  * Added window utilities: get_data_window(), window_union(),
    window_intersection(), windows_intersect().
  * Warn when an alpha band that might provide a dataset mask is shadowed by a
    nodata attribute.
  * IPython is not the default interpreter for rio-insp and the documentation 
    saying it is has been corrected.
  * Guard against creating datasets with block sizes larger than the dataset
    width and height. Such datasets are semi-broken and are likely to be 
    mangled when read.
  * Refactor of the `rasterio.features` tests.
- Update to version 0.29.0
  * Fill masked arrays in rio-calc when using Numpy 1.10.x as well as with 1.8.x.
  * When a raster dataset is not tiled, blockxsize and blockysize items are no
    longer included in its `profile` property. This prevents meaningless block
    size parameters from stripped, not tiled, datasets from being used when
    creating new datasets.
- Update to version 0.28.0
  * Ensure that tools module is packaged. The rio-merge command was
    broken in 0.27.0 and is restored to working order in version 0.28.0.
  * Add `precision` keyword argument to `index()` method.
- Update to version 0.27.0
  * Ensure local uniqueness of the rio-shapes feature ids.
  * Surface compression and interleaving as dataset properties and in rio-info.
    In the module, these are enums (`enums.Compression` and 
    `enums.Interleaving`); the values of the enums correspond to GDAL terms
    (i.e, "DEFLATE") and the names are what surface in the CLI ("deflate").
  * Change get_window() and DatasetReader.window() to return a window guaranteed
    to cover the input bounding box.
  * Bug fix for improperly computed transforms of output file in tools.merge and
    rio-merge.
  * More robust determination of dataset nodata values. In particular, the 
    absence of a nodata value is much more clear: dataset.nodata should never
    return an out of range value when there is no nodata value, it should always
    return `None`.
- Update to version 0.26.0
  * Add dependency on click-plugins, a new project that takes over the plugin
    duties formerly assigned to cligj.
  * Change rio-stack's --photometric=RGB option to --rgb. Other 
    photometric interpretations should be assigned using the --co option 
    added in 0.25.0.
  * Allow for ndarray-like objects (like xray arrays), not just numpy arrays,
    in warp() and elsewhere.
  * Add --rgb flag to rio-convert.
  * Fixed resampling algorithm enumeration bug.
  * Colormap handling was made too strict in 0.24.1 and has been made more
    forgiving. Callers are now warned when alpha values will be ignored instead
    of receiving exceptions.
  * Add a .gitignore. Better late than never!
  * Add a checksum() method to base dataset class and checksums to output of
    rio-info.
- Update to version 0.25.0
  * New rio-warp command
  * Add driver-specific creation options (`--co`) to many commands.
  * Add support for arbitrary CRS output to rio-bounds.
  * Add support for getting values from template files in rio-edit-info with a
    `--like` option.
  * New rio-overview command.
  * Fix rounding error in extracting shapes from decimated data.
  * Remove creation options from meta property and move them to new profile
    property.
  * Fix for bug in passing affine keyword argument to open() in 'w' mode.
  * New rio-convert command, a replacement for gdal_translate 
    with more features to come by 1.0.
  * Improved error messages when seeking a driver when none are registered.
  * Replace read_band() with read() in the rio-insp banner.
  * Fix an indexing error that prevented window() and window_bounds() from 
    round-tripping properly.
- Update to version 0.24.1
  * Improve safety of the sample() generator.
  * Provide array masking features missing from Numpy<1.9.
  * Guard against attempts to write RGBA colormap entries to TIFFs, which the
    format can not support.
- Implement single-spec version
- Switch to github sources which include test data

-------------------------------------------------------------------
Tue Jun 09 12:34:00 UTC 2015 - Angelos Tzotsos <tzotsos@opensuse.org>

- initial packaging (0.24.0)

openSUSE Build Service is sponsored by