libdap-3.21.1 for Hyrax-1.17.1

libdap-3.21.1

News for libdap-3.21.1

• Initial support for UTF-8
• Improved production rules.
• Worked technical debt affecting large variables and large responses (large >= 2Gb).

libdap-3.21.0-27 for Hyrax-1.17.0

News for version 3.21.0-27

• Added new Direct I/O support so that modules written using libdap
  can pass compressed data buffers read directly from disk to output
  files without expensive decompression and recompression operations.
• Merged contributed fixes from Bo Anderson
  Fix handling of libtirpc pkg-config files with -L flags (#228)
• Merged contributed fixes from Dan Horák <da at @danny.cz>
  add missing include (#227)
  With GCC 13 the header isn't included thru other headers any
  more, thus include it explicitly. Otherwise uint8_t or uint32_t type
  remain undefined in Vector.cc.
  Fixes: #226
  add missing big endian baselines (#196)
• Fixed a bug where the copy ctor for D4Maps failed to correctly set the parent Array.
• Merged contributed fix from Orion Poplawski
  Add missing cstdint include for uint32_t (#219)
• Removed support for RHEL 7 (CentOS 7)
• Now require C++-11 to build the code. However, configure will use
  C++-14 if it finds that and the next release will require that.
• Moved the functionality of is_dap4_projected() into libdap4 (#213). This
  had a number of consequences and there are new methods to support the feature.
• Added support for 64-bit sized arrays. This was done by adding a set of
  'size methods' that have the suffix '_ll' (for long long). These should
  be used in place of the old methods, which are still in the code.

libdap 3.20.11 for Hyrax 1.16.8

• Fixed bug in computation of request_size_kb()
• Fixed type issue in HTTPCache.cc (#192)

libdap-3.20.10

• Support for RHEL8
• Fix for bugs in the ce parser around inverted indices.
• Fix for libdap4 github issue 147: Grid::get_map_iter() was off by one
• Improvements to DAP4 api.
• Fixed various memory leaks.
• Replaced instances of &vector[0] with vector.data() (RHEL8)

libdap-3.20.9

Updates in support of Hyrax-1.16.5

• Started migrating from (deprecated) auto_ptr to C++11 unique_ptr

• Migrated use of regex functions from outdfated GNU implementation
  to C++11 implementation (uses compile time swicth)

libdap 3.20.8 for Hyrax 1.16.4

Modified Error so that it is more in line with C++11.

Added (shallow) unit64_t request size computations and max_request_size state.

Changed the internal representation of max response size to uint64_t and confined the overflow
to just the (deprecated) functions.

libdap-3.20.7 for hyrax-1.16.3

In this minor release of libdap we:

• Added code coverage (using gcov) to the CI processes.
• Dropped support for CentOS-6 (whew!)
• Improved Int64 support.
• Corrected byte order issue with DAP4 data transmission.
• Improved error reporting.

libdap-3.20.6 for hyrax-1.16.2

This minor release is part of the Hyrax 1.16.2 minor release.

Updates:

• Stopped CE parse errors from returning user supplied strings in error messages.
• Switched to README.md from README
• Swapped out unique_ptr for auto_ptr because of CentOS6 issues.

Version 3.20.5 for Hyrax 1.16.1

News for version 3.20.5

Fixed memory leaks in imported aws signing code.

Bug in MDS in which LMT was not being utilized.

Fixed additional memory leaks in the libxml2-based

Lots of continuous integration work.

Version 3.20.4 for Hyrax 1.16.0

News for version 3.20.4

Fixed memory leaks in the libxml2-based DMR and DDX parsers.

Moved more toward C++-11. On CentOS 6 we now build using
c++0x while CentOS 7 uses c++11. This means that we can
use unique_ptr<> but not null_ptr, for example.

We have found a bug in the code that orders Dimensions,
Enumeration definitions and Groups in the DMR. Listing the
Groups last makes DMRs that reference Dimensions defined inside
Groups fail silently.