Age | Commit message (Collapse) | Author |
|
Do not read and write data blocks in whole chunks. Read and
write value by value and do the md5 summing in steps.
This hopefully avoids memory alignment problems reported
on ARM architectures.
Fixed a bug freeing trajectory mapping memory.
Change-Id: I3438b3d3faf5349dcd477a31cfb4bacf236ed582
|
|
Added functions to get the version of the header and the
library.
Improved version control of the library using CMake config and
version files.
Changed tng_util_frame_current_compression_get() return type
to double before the API is widely circulated (no release
has been made since that function was added).
Compiling compression functions into the tng_io library.
Uses BuildTNG.cmake instead of including add_subdirectories.
Change-Id: I5e4818736fb2632c06ca7df8e84ce9544bf2acaf
|
|
TNG_VERSION was renamed to TNG_API_VERSION which is configured
through API_VERSION in CMakeLists.txt.
Updated the header to 2014.
Change-Id: I6a37825cd1d236927ac0e21a797abc4e98f7ce59
|
|
|
|
|
|
|
|
|
|
Explicitly compare signed and unsigned in many places to avoid
warnings.
|
|
Minor fixes.
|
|
|
|
Also make compression slightly faster.
|
|
|
|
Reading box shape using the high-level API should work.
Fixed bugs in data reading.
Updated utility usage examples.
|
|
|
|
Fixed many bugs in the high-level API.
Use the high-level API (where appropriate) in new example files.
|