Releases: glotzerlab/gsd
Releases · glotzerlab/gsd
v2.0.0
Note
- This release introduces a new file storage format.
- GSD >= 2.0 can read and write to files created by GSD 1.x.
- Files created or upgraded by GSD >= 2.0 can not be opened by GSD <
1.x.
Added
- The
upgrade
method converts a GSD 1.0 file to a GSD 2.0 file in
place. - Support arbitrarily long chunk names (only in GSD 2.0 files).
Changed
gsd.fl.open
acceptsNone
forapplication
,schema
, and
schema_version
when opening files for reading.- Improve read latency when accessing files with thousands of chunk
names in a frame (only for GSD 2.0 files). - Buffer small writes to improve write performance.
- Improve performance and reduce memory usage in read/write modes
('rb+', 'wb+' and ('xb+'). - C API: functions return error codes from the
gsd_error
enum.
v2.x integer error codes differ from v1.x, use the enum to check.
For example:if (retval == GSD_ERROR_IO)
. - Python, Cython, and C code must follow strict style guidelines.
Removed
gsd.fl.create
- usegsd.fl.open
.gsd.hoomd.create
- usegsd.hoomd.open
.GSDFile
v1.0 compatibility mode - usegsd.fl.open
.hoomdxml2gsd.py
.
Fixed
- Allow more than 127 data chunk names in a single GSD file.