2.18.3 2017-09-22 ~~~~~~~~~~~~~~~~~ * The CODA Python interface now automatically sets a default CODA_DEFINITION path relative to the CODA Python package. * Fixed memory leak in HDF5 backend. * Allow fixed length array dimension definitions for XML formatted products. 2.18.2 2017-06-20 ~~~~~~~~~~~~~~~~~ * Added automatic scale_factor/add_offset/missing_value/_FillValue conversion for HDF5 (similar to what was already done for netcdf and HDF4). * Leading and trailing whitespace that is allowed when parsing numerical values from ascii data can now consist of either space or a tab characters (instead of just space characters). * Improved reading performance of GRIB messages that use 'simple packing'. 2.18.1 2017-03-23 ~~~~~~~~~~~~~~~~~ * Fixed issue with reading of XML files using a codadef definition when mmap is disabled. * Fixed issue where asciiline length of last line was 1 byte too short if line did not end with an end-of-line. * Improved Python coda.fetch() performance. 2.18 2017-01-20 ~~~~~~~~~~~~~~~ * Changed Open Source licence of CODA from GPLv2 to BSD (3-Clause). * Added support for using path references as a parameter to the CODA fetch() functions in the Python, Matlab, and IDL interfaces. You can now do things like 'coda_fetch(pf, "dataset[0]/variable@unit")'. * CODA now supports GRIB files that contain a mix of GRIB1 and GRIB2 messages. GRIB files are now also indicated as plain GRIB (coda_format_grib) and no longer as either GRIB1 (coda_format_grib1) or GRIB2 (coda_format_grib2). These changes are backward compatibility breaking. If you read GRIB data you will have to adapt to the new mapping. * Added listOfNumbers field to GRIB mapping which provides the list of numbers of points in each row for quasi-regular grids. * Fixed issue where fields of a union were not marked as optional. 2.17.3 2016-09-30 ~~~~~~~~~~~~~~~~~ * Fix in autotools build where coda executables were incorrectly linking to shared coda library if --disable-static was provided to configure script. * Fix issue where in a specific case (definition had no attributes and xml element had a namespace) no error was raised for an unallowed xml attribute. * Added entry for 2016-12-31 in leap second table. * Fixed issue in codadump where for products with a self describing format it was using the available codadef definition (if available) to navigate the product, instead of using the actual (self described) format. 2.17.2 2016-09-02 ~~~~~~~~~~~~~~~~~ * Fix compatibility issue with different flex versions for autotools build. 2.17.1 2016-08-25 ~~~~~~~~~~~~~~~~~ * The open_as() function can now force a file using a self-describing format (e.g. HDF5, XML, etc.) to be opened as a raw binary/ascii file if the product definition is binary/ascii. * Fixed open file handle counting leak in IDL interface when coda_open() failed. * Fixed enabling of HDF4/HDF5 in CMake build on Windows. * Allow xml attributes with a namespace in xml products to be defined in a codadef by a definition of an attribute without a namespace (as was already possible for xml elements). * Fixed wrong types for fixed surface fields in format mapping documentation for GRIB format. 2.17 2016-06-27 ~~~~~~~~~~~~~~~ * CMake build system can now be used on Windows. It currently only supports building of the C library, command line tools and Python interface. * Updated embedded PCRE library to version 8.38. * Updated embedded expat library to version 2.1.1. * Updated embedded zlib library to version 1.2.8. 2.16 2016-03-24 ~~~~~~~~~~~~~~~ * Automatic detection of product class/type/version is now performed using CODA expressions and is now possible for all data formats. All existing detection rules in codadefs are automatically translated to the new CODA expression variants when the codadefs are read. codadefs that tried to map hdf4, hdf5, cdf, netcdf, grib1, grib2, rinex, or sp3 files to ascii/binary to allow automatic product type/version detection will have to be updated to now perform the detection using the actual format of the product. * Product definitions in codadefs may now omit a root type definition if the product is not an 'ascii' or 'binary' product. The root type will then be determined purely dynamically based on the content of the file. This allows for codadef entries that are purely meant to detect the product type/version for a product using a self-describing data format. * codacheck can now verify the format of self-describing formats such as netCDF, HDF4, HDF5, CDF, if a definition is provided in a codadef file. * Added coda_expression_is_equal(), coda_expression_print(), and coda_cursor_print_path() functions. * Added 'at', 'dim', and 'numdims' coda expression functions. * Introduced several additions to the codadef format in order to ease the transition to a future revision of the codadef format: - added MatchExpression - allow Attribute to be used for Array, Union, Integer, Float, Text, Complex, Raw, and Time - introduced real_name attribute for Field and Attribute (which is mutually exclusive with the use of xmlname for types) - allow Attribute to contain a type definition (similar to Field) if an explicit type definition is used then any FixedValue should be provided as part of the type (and not as part of Attribute). - allow Hidden and Available to be used for Attribute (similar to Field) * CODA now stores the namespace of an xml element in an attribute field called 'xmlns'. This only happens for xml products that have no codadef definition or, if there is a codadef, where the codadef definition explicitly allows the (optional) presence of an 'xmlns' attribute. * XML products with an xml namespace will now be matched by codadef detection rules that do not include a namespace in the path matching rule. Only rules that explicitly check for an xml namespace will now restict matches to only those xml products that use that same xml namespace. * Fixed issue in list/ascii/hdf4 modes of codadump where field availability was not correctly determined if the availability of the field would differ per array element. * All exported symbols of the CODA shared library now have a 'coda_' prefix. * Fixed memory corruption issue in evaluating coda expressions that used a string constant containing the '\0' character. * Fixed double definition problem in coda.h for MinGW. * Fortran coda.inc include file is now a plain source file and is no longer generated using the generate-finc tool. The generate-finc tool has also been removed (which improves cross-compilation). 2.15.1 2015-12-09 ~~~~~~~~~~~~~~~~~ * Fixed memory issue in Python cursor_read_complex_* functions. (SWIG wrapping was using complex64 where it should have been complex128). * Product variables are now explicitly set to 0 before the initialization expression is evaluated (eliminating undefined behaviour when the init expression was somehow trying to use the value of the variable itself). * Windows build now depends on HDF4 4.2.11 and HDF5 1.8.16. * Added better clarification in the C Library documentation on the epoch choice for the leap-second aware time conversions. 2.15 2015-11-03 ~~~~~~~~~~~~~~~ * Removed QIAP support as it is no longer going to be a supported standard. * Added --max_depth option to codadump debug mode. * Added bytes(node, offset, length) coda expression function. * Resolved possible windows build issue. Now using CreateDirectory() instead of mkdir(). * Changed detection block size from 4096 to 16400 bytes (allowing for detection of HDF5 files where the superblock starts at offset 4096, 8192, or 16384). * Improved CMake build system. * Removed stdio.h #include from coda.h * Fixed name of string cast function in generated codadef documentation ('str' instead of 'string'). * Resolved issue with FLEX generated files not properly prefixing all exported symbols. * Fixed memory leaks in different backends (introduced in 2.14). 2.14 2015-09-04 ~~~~~~~~~~~~~~~ * HDF4 character array attributes are now interpreted as strings. Note that this breaks backward compatibility for any routines that read attribute data from HDF4 files using CODA. * Improved support for reading of HDF5 attributes: - Scalar attributes are no longer inside an array element - Allow reading of scalar variable length string attributes * Added partial (offset+length) array reading functions to C, Python, Fortran, and Java interfaces. * Added JSON and YAML export options to codadump. * codaeval now properly throws an error when trying to evaluate a 'void' or 'node' expression. * Fixed sorting of product classes (last product class was not allways included in the sorting). * Fixed memory issue in gracefull handling of HDF5 files with invalid content. * Fixed issue where coda_cursor_goto() could navigate to the wrong record field in case there was a field available whose name started with that of another field in the same record. * Added 'definition' mode to codadd tool. * MATLAB MEX extensions on Unix-based platforms are no longer build using the 'mex' script that comes with MATLAB, but using self derived compilation settings. * Dropped support for MATLAB R11/R12 and IDL V5.x. 2.13 2015-04-29 ~~~~~~~~~~~~~~~ * Fixed Python wrapping of functions coda.time_string_to_parts, coda.time_string_to_double, and coda.time_string_to_double_utc. (they now accept both format and string parameters again) * Added Python 3 compatibility. In Python 3 coda_expression_eval_string() returns 'bytes' and coda_cursor_read_string() returns 'unicode'. * Added --open_as option to codadump debug mode. * Doxygen documentation is now made with doxygen 2.8.9.1. 2.12.1 2015-02-24 ~~~~~~~~~~~~~~~~~ * Added option to 'codadump debug' to disable the use of fast-size expressions. * offset and size arguments to Python coda.cursor_read_bytes() function are now optional. * Fixed issue where strings returned via the Python interface were sometimes shortened if the string contained '\0' or other non-printable characters. * Fixed issues where detection of xml files could fail. Problems could occur when an element was fully empty (not even containing whitespace) or when the element string content or closing tag crossed an 8Kb boundary in the file. * Added entry for 2015-06-30 in leap second table. 2.12 2015-01-15 ~~~~~~~~~~~~~~~ * Fixed memory corruption in 'time to string' functions in Python interface. * Added -D option to all command line tools to allow explicit setting of CODA definition path. * Added feature to reference record fields by index in CODA expressions using the '/{0}' style syntax. * Improved performance of reading vsf_integer data. 2.11.1 2014-11-14 ~~~~~~~~~~~~~~~~~ * Detection of HDF5 files now also supports HDF5 files where the superblock is not located at the start of the file but at offset 512, 1024, or 2048. Files that have the superblock starting at higher offsets are still not supported. * Error messages for offset/size/availability expressions now show at which cursor position the error occured. * Updates to windows build systems. - Fixed building against latest version of HDF5 (hdf5-1.8.13). - CODA binaries are now built against Python 2.7.8 and numpy 1.8.2. - Microsoft Visual C++ runtime libraries are no longer included with the Windows binary installer. You now have to install these yourself: http://www.microsoft.com/en-us/download/details.aspx?id=29 http://www.microsoft.com/en-pk/download/details.aspx?id=15336 2.11 2014-08-27 ~~~~~~~~~~~~~~~ * Refactored XML backend and memory backend All XML data is now first fully read in memory (using expat) and then accessed from there. Since CODA no longer tries to directly read from the files anymore this means that: - all data will now be correctly returned as UTF-8 (even if the XML file had a different encoding) - XML entities are now properly parsed - error messages that quote xml element content will no longer display parts of the xml closing tag in case the xml content was shorter than expected. Other backends that used in-memory storage (e.g. SP3, RINEX, GRIB, CDF) also have been refactored to use the new memory backend. * SP3, RINEX, CDF and GRIB backends now present numbers more accurately (i.e. matching the underlying float or double precission). * Reworked all CODA time functions. The following functions have been introduced: coda_time_double_to_parts(), coda_time_double_to_parts_utc(), coda_time_parts_to_double(), coda_time_parts_to_double_utc(), coda_time_parts_to_string(), coda_time_string_to_parts(), coda_time_double_to_string(), coda_time_double_to_string_utc(), coda_time_string_to_double(), coda_time_string_to_double_utc(). The following functions are now deprecated and will be removed in a future CODA release: coda_datetime_to_double(), coda_double_to_datetime(), coda_time_to_string(), coda_string_to_time(), coda_utcdatetime_to_double(), coda_double_to_utcdatetime(), coda_time_to_utcstring(), coda_utcstring_to_time() Refer to the bottom section of the libcoda/coda-time.c source code to see how to map from the old functions to the new functions. * The CODA expressions time format now supports '*' as a special character to indicate that values should have leading spaces instead of leading zeros. * Improved debug output of codadump * Added --no-mmap option to codacheck * Improved Fortran interface documentation. Default configuration and examples now use 64-bit interface instead of 32-bit interface. * Changed default Python version for Windows build to Python 2.7. 2.10.1 2014-03-21 ~~~~~~~~~~~~~~~~~ * Fixed use of CODA_BUILD_SUBPACKAGE_MODE in CMake build. * Fixed small html documentation generation issue in codadd for conversions. 2.10 2014-02-01 ~~~~~~~~~~~~~~~ * Added support for 'offset' and 'invalid' attributes for a Conversion element in codadef files. * Added int(bool) to CODA expressions. * Added size check for content of xml elements to codacheck. * Fixed issue where coda_cursor_has_attributes() and coda_type_has_attributes() would always return false for HDF5. This could result on attributes not being shown in several interfaces (e.g. codadump) * Fixed bug in codadump debug argument parsing. 2.9.1 2013-06-27 ~~~~~~~~~~~~~~~~ * When parsing an ascii datetime value fails, codacheck will now continue checking the remainder of the product. 2.9 2013-05-22 ~~~~~~~~~~~~~~ * Added 'SwapDimensions' option to IDL and MATLAB interfaces (default enabled). With this option you can disable the automatic transpose operation that CODA performs on multi-dimensional array data. 2.8.1 2013-03-07 ~~~~~~~~~~~~~~~~ * Fixed issue where ascii time value format definitions with an ascii mapping that maps the empty string to NaN, had this mapping applied for any value (regression introduced in 2.8). * Fixed issue where trying to use coda_cursor_read_bytes/bits on a 'no data' special type element could cause a segmentation fault. * Time parsing of a string now allows the fractional part of a second to be longer than 6 digits (although remaining digits will be ignored and the time value will always be truncated at microsecond resolution). * The 'complex' special type is no longer restricted to binary data (but can now also be used for e.g. ascii data formats). 2.8 2012-12-12 ~~~~~~~~~~~~~~ * Added coda_strfl() to the C interface to support platform independent printing of nan/inf floating point values. * Added time() and strtime() to CODA expressions. * Product data of type 'time' now uses CODA expressions to calculate the actual date/time value. * coda expression 'str()' can now also turn integer values into strings. * Power operator in CODA expressions (i.e. 'x ^ y') where both operands are integers will now return a float instead of an integer. Also fixed issue in case second operand was a negative number. * Floating point CODA expression result values are now printed in exponential format for large numbers (instead of always using fixed point format) and use a standardised way of printing 'nan' and 'inf'. * codacmp now has an option to compare array elements by aligning them using a 'key expression' (--key option). * Added --path option to codadump debug mode. * Renamed 'pv' field in GRIB1 products to 'coordinateValues'. Added 'coordinateValues' field for GRIB2 products (for vertical coordinates). 2.7.2 2012-09-24 ~~~~~~~~~~~~~~~~ * Fixed broken compilation from source on big-endian systems. 2.7.1 2012-09-18 ~~~~~~~~~~~~~~~~ * Fixed issue where codadef definitions (for product type detection) for HDF4, HDF5, CDF, netCDF, and GRIB formats were not properly associated to an open product. * Fixed issue where the global option bypass_special_types was not handled properly by non-ascii/binary backends (e.g. SP3, RINEX, CDF). * Fixed issue with CDF backend where variables for which the 'varies' property for all dimensions (including the record dimension) was False resulted in an error. These variables are now treated properly as scalar variables. 2.7 2012-08-31 ~~~~~~~~~~~~~~ * Added support for reading of CDF data (using direct access to the CDF files, so the CDF library is not needed). Note that not all features and versions of CDF are supported. See the CODA documentation for details. * Fixed issue in codadump debug mode where arrays of time values only showed the time value for the first array element (and only the base type value for the rest). * Fixed issue where QIAP correction actions were not properly applied when calling coda_cursor_read_*_array() functions. * Several improvements to build system of CODA Python interface: - fixed building of CODA Python interface with CMake. - all CODA Python interface files are now installed in the platform specific python library directory. On 64-bit Linux systems this means that the full CODA python package ends up in /lib64/site-packages/coda. - improved autoconf and CMake based detection of default numpy include directory. * The binary backend now also supports data of type text (= char or string). Previously such data always had to use the 'ascii' format. 2.6 2012-07-07 ~~~~~~~~~~~~~~ * Fixed issue in codacmp where strings of unequal length were not indicated as a change. * Added support for detection of xml files (as defined in the codadef files) based on xml attribute value and optionally additionally on filename. Detection order for xml has been changed, it is now performed 'depth first'. * Added -d option to codacheck: produces an error if no codadef entry for a product was found. * Enabled building of the CODA Python interface using the CMake build. * Fixed cases where command line tools could send error messages to stdout instead of to stderr. 2.5 2012-05-14 ~~~~~~~~~~~~~~ * Improved support for GRIB2 data: - fixed reading of Data Representation Section information - added fields for Product Definition Section * codacheck now checks for expected end of line sequences in ASCII files again. * Fixed issue where truncated blank characters in RINEX Observation records were not properly supported. 2.4.1 2012-04-18 ~~~~~~~~~~~~~~~~ * Fixed handling of nan and inf time values in the IDL interface when the TimeUnitDays option was enabled. * Filenames of product definition html documentation files from 'codadd doc' are now based on product definition name instead of product type + version. * Fixed problem with compilation of Fortran interface. 2.4 2012-02-23 ~~~~~~~~~~~~~~ * Added coda_cursor_goto(), coda_type_has_attributes(), and coda_cursor_has_attributes(). * Added basic support for RINEX Clock (2.0 and 3.0) format. * Added 'with' function to CODA expression language and introduced additional index variables ('j' and 'k') for the 'for' and 'with' functions. * Improved document to better explain setting of CODA definition path. * Added missing utc datetime functions to Fortran interface. * Fixed issue in printing of attributes for 'codadd list'. * Fixed memory issues with MATLAB interface on 64-bit systems. 2.3.1 2012-01-21 ~~~~~~~~~~~~~~~~ * Added leap second for 2012-07-01 to default leap second table. * CODA IDL function CODA_SIZE now returns '0' when you reference a scalar. * Fixed installation paths on Unix in CMake build. * Fixed issue where XML attribute names when accessing plain XML files (without codadef) could erroneously get a numeric postix. 2.3 2011-12-13 ~~~~~~~~~~~~~~ * Improved SP3 support - Extended support to SP3-a and SP3-b. - CODA format enumeration value is now called 'sp3' instead of 'sp3c'. - Fixed issue where SP3 products < 2000 and >= 2010 were not supported. * Added support for QIAP (Quality Information and Action Protocol). See libcoda/qiap/README file in source package for details. * Added coda_expression_get_type_name(). * Fixed the format of parameter paths in the output of codacheck and codacmp. * Added missing CODA expression interface for Fortran. * Added --parent-types and --attributes option to 'codadd list'. * Configuring the build system for the Java interface on Mac OS X should now work correctly. * Fixed issue where CODA would no longer create separate record fields for netCDF elements and HDF4 root elements (vdata, vgroup, etc.) if elements shared the same name (in which case CODA should append unique _## numeric postpixes to the fieldnames). * Fixed potential wrong reading of epoch value from RINEX Navigation record. * Fixed memory leaks in netCDF, GRIB, and RINEX backends. * Fixed issue where CODA did not properly support HDF4 dataset names that are longer than 63 characters. * Upgraded the Windows Visual Studio 6.0 project files to Visual Studio 2008 (9.0) project files. * The binary installer of CODA for Windows now requires IDL 6.3 or higher. 2.2 2011-09-02 ~~~~~~~~~~~~~~ * Added coda_open_as(). * Fixed issue where CODA was producing error messages when trying to read Variable Scale Factor integers. * Fixed endian conversion problem in GRIB backend for Grid Definition Section data. * Fixed issue where MATLAB coda_size() function did not work on the root of a product if the root was an array. * Fixed issue where expressions on ascii content in xml products were not able to use the attribute content of the encapsulating xml element. * Fixed matching of product format detection expressions based on file names on Windows when the path to the file used '\' characters instead of '/'. * libcoda is now linked statically into the CODA IDL DLM again on Unix systems (regression from CODA 1.1). This eliminates the need for users to set a runtime shared library path. * Fixed wrong fieldname in CODA documentation of SP3-c mapping. * Removed the text format restriction on 'product type' names (they no longer have to consist of only a-z, A-Z, 0-9 and '_' characters). 2.1 2011-07-21 ~~~~~~~~~~~~~~ * CODA now has a logo. * Added coda_type_get_attributes(). * Added support for SP3-c format. * Added XML Schema export option to codadd. * Fixed documentation for 'exists()' expression. This function can also be used on non-existing paths (instead of only being able to test for availability of optional record fields). 2.0.1 2011-05-17 ~~~~~~~~~~~~~~~~ * Fixed issue that prevented correctly opening GRIB files. 2.0 2011-04-20 ~~~~~~~~~~~~~~ * Major redesign of the backend architecture of CODA with the aim to: - Easier add format-specific backends for which the data will be kept fully in memory, such as for the RINEX and SP3c formats. - Prepare for future support for specification of data formats through .codadef files for self describing formats such as HDF5 and netCDF for format verification purposes. * Added RINEX backend, currently only supporting Observation and Navigation data for RINEX version 3.0. * Added skeleton for upcoming SP3-c backend. Although SP3-c files can currently be opened, the product is mapped as an empty record. A full implementation of the SP3-c backend will arrive in the next CODA version. * Removed coda_type_has_ascii_content() function. It is now only possible to determine the 'ascii content' property dynamically using coda_cursor_has_ascii_content(). * Added coda_eval() function to IDL and MATLAB interfaces to allow evaluation of CODA expressions. * coda_type_get_record_field_real_name() now gives the actual name for: - XML element and attribute names - HDF4 GRImage, SDS, Vdata, and Vgroup names - netCDF variable and attribute names * Added coda_type_get_record_field_index_from_real_name() * Added productformat() to CODA expressions. * Improved performance of reading bitmasked GRIB data and fixed memory leak in GRIB backend. 1.5 2010-12-20 ~~~~~~~~~~~~~~ * Added coda_get_product_definition_file(). * Added support for Bit Maps for GRIB1 and GRIB2. * Fixed possible memory corruption when coda_init() was called again after coda_done(). * Fixed memory leaks in GRIB backend for coda_done(). * Fixes in CMake build. 1.4.1 2010-12-01 ~~~~~~~~~~~~~~~~ * CODA can now be build using MSYS/MinGW on Windows. * Fixed issue with GRIB backend on systems that require strict word alignment. * Added support for GRIB files with spacing between the GRIB messages. 1.4.0 2010-11-08 ~~~~~~~~~~~~~~~~ * Added support for reading of GRIB1 and GRIB2 formatted data. Support is currently limited to data stored with simple packing and using a lat/lon or Gaussian grid. * Added public C API for creating and evaluating CODA expressions. A wrapper interface is available for Fortran, Python, and Java. * Added regex functions to CODA expression language. The functions use PCRE, which is now included with CODA, to evaluate the regular expressions. * Added support for raw strings (i.e. strings where the escape sequences are not evaluated by CODA) in the expression language. A raw string starts with 'r' (similar to Python): r"no \\escaping\\" -> "no \\escaping\\" * Added codaeval command line tool to verify and evaluate CODA expressions. * Added new time functions to the C interface (with Fortran, Python, and Java wrappers) that can deal with leap seconds (they convert between TAI and UTC): coda_utcdatetime_to_double(), coda_double_to_utcdatetime(), coda_time_to_utcstring(), coda_utcstring_to_time() * C interface types coda_ProductFile, coda_Type, and coda_Cursor are now called coda_product, coda_type, and coda_cursor. Defines are still available for the old type names to allow backward compatibility. * Added coda_type_get_record_field_real_name(). * Added coda_set_definition_path_conditional(). * codadump no longer shows execution time information for HDF4 export. The list view of codadump also does not quote the type names anymore. * Fixed error handling of Python coda.string_to_time() function. * Added variant of unboundindex() expression function that can take an explicit termination condition as third argument. * Windows build of CODA is now performed using Visual Studio 2008 (9.0). The MATLAB build on Windows now requires MATLAB R2007b or later. 1.3.6 2010-06-11 ~~~~~~~~~~~~~~~~ * Fixes in CMake build files for big-endian systems. 1.3.5 2010-05-10 ~~~~~~~~~~~~~~~~ * Added support for reading of 24, 40, 48, and 56 bit little endian integers. 1.3.4 2010-04-02 ~~~~~~~~~~~~~~~~ * Fixes in CMake build files. 1.3.3 2010-03-19 ~~~~~~~~~~~~~~~~ * Fixed small issue in html documentation generation (wrong byte end-position for a detection expression). * Fixed bug in parsing of string constants in expression language (string constants can now end with an escaped backslash '\\'). 1.3.2 2010-01-08 ~~~~~~~~~~~~~~~~ * NetCDF backend now attaches attributes to the array type instead of the array element type for netCDF variables. * Fixed memory leaks in HDF5 backend. * Special SDSVar and CoordVar SDSs are now properly hidden when accessing HDF4 products. * Error handling was improved for a few special cases. 1.3.1 2009-09-25 ~~~~~~~~~~~~~~~~ * Added Java interface. * CODA expression 'length' now also works on 'node' expressions. * Fixed behavior of coda_cursor_read_bits() when the bit_length argument is 0. * Added support for reading of arrays of basic types for XML products where the array is not an ASCII (space separated) array, but each ASCII value is stored in a separate XML element. 1.3.0 2009-07-17 ~~~~~~~~~~~~~~~~ * Added built-in netCDF backend for accessing netCDF3 32-bit and 64-bit data formats (reading of netCDF data is no longer performed via the HDF4 library). * scale_factor/add_offset conversion for HDF4 SDS data is now also applied if the scale_factor and/or add_offset attributes are provided as FLOAT32 (and not only if they are provided as FLOAT64). * Fixed coda.string_to_time() return value mapping in Python interface. * Fixed issue with parsing of microsecond information from ascii EPS time string values. * Fixed fetching of array of raw byte data in IDL interface. 1.2.3 2009-06-10 ~~~~~~~~~~~~~~~~ * coda_get_product_format() now properly returns coda_format_ascii for ascii files. * Removed -t (totals) option of codacheck. * Added coda_str64() and coda_str64u() to the C interface to support printing of 64 bit values on 32-bit platforms where printf does not support %lld. * Fixed issue that could prevent CODA from working correctly on systems that are picky about having variables aligned on 32 or 64 memory boundaries. 1.2.2 2009-05-08 ~~~~~~~~~~~~~~~~ * CODA expression 'if' now also works with boolean typed 2nd and 3rd argument. * CODA expression to read 'raw' data now properly reads the number of required bits (and not any padding bits that could be in the sequence of the rounded up number of bytes). The same issue was also fixed in codacheck and codadump. * codacheck now checks the file size of a product in number of bits instead of rounded up number of bytes (i.e. undefined spare bits at the end of a product is now considered an error). * Binary signed integers in CODA definitions are no longer limited to a multiple of 8 bits. * It is now possible to define 'fixed values' in format definitions for data that is not a rounded number of bytes. * Solved several compiler warnings regarding 64-bit support. * codacmp now escapes 'unprintable characters' when printing differences in 'raw' data. * Fixed name of coda.get_product_format() in Python interface. 1.2.1 2009-03-01 ~~~~~~~~~~~~~~~~ * Fixed '|' and '&' usage in CODA expressions. 1.2.0 2009-02-20 ~~~~~~~~~~~~~~~~ * The Python interface of CODA now uses numpy instead of numarray. * The codadd command line tool now has a 'list' mode to inspect available format definitions. * Fixed support for HDF5 datasets containing variable length strings. * Added support for coda_format_netcdf and coda_format_cdf as official formats. CODA will however (still) use the HDF4 backend to read data from these files. Note that coda_format_netcdf only referes to pre-4.0 NetCDF formatted files. NetCDF 4.0 files are in HDF5 format and will thus have coda_format_hdf5. * Fixed issue with reading of rank-0 SDS data from HDF4 and NetCDF files. * Fixed issue where all fieldnames in HDF4 files got called 'unnamed' and where reading of VData data failed. * Changed maximum cursor depth (CODA_CURSOR_MAXDEPTH) from 16 to 32. * Added ceil(), floor(), round(), isnan(), isinf(), ismininf(), isplusinf(), ltrim(), rtrim(), trim(), and unboundindex() functions and bitwise AND '&' and OR '|' operators to CODA expressions. * CODA expressions now accept floating point constants in FORTRAN format (using a 'D' for specifying the exponent: e.g. .133000D+03) * Various improvements to reading of .codadef files. 1.1.1 2008-10-21 ~~~~~~~~~~~~~~~~ * The required zlib files are now included in the CODA package and all zlib functions are 'name mangled'. This solves an issue with running the CODA IDL interface on Linux. 1.1.0 2008-10-10 ~~~~~~~~~~~~~~~~ * Fixed the following CODA expression operations - multiplication of floating point values - reading of floating point values * CODA expressions now allow combining integer and floating point arguments in a floating point expression (e.g. compare integer against float, or add float to integer) without the need to explicitly cast the integer to a float. * Added abs() function to CODA expressions. * Upgraded to expat 2.0.1. * Files with an ASCII definition are now properly marked as files with an ASCII format. 1.0.0 2008-09-24 ~~~~~~~~~~~~~~~~ * First official release.