Releases: esi-neuroscience/syncopy
Releases · esi-neuroscience/syncopy
v0.1a1
[v0.1a1] - 2019-10-14
Preview alpha release of SyNCoPy for first ESI-internal tryout workshop.
NEW
- Added routines
esi_cluster_setup
andcluster_cleanup
to facilitate
using SLURM from within SyNCoPy - Included new
FauxTrial
class and_preview_trial
class methods to
permit quick and performant compute dry-runs - Included a
select
keyword to allow for in-place selections that are applied
on the fly in any meta-function via a new decorator. The heavy lifting is performed
by a newSelector
class - Re-worked the
specest
package:mtmfft
is now fully functional - Overhauled HTML documentation
CHANGED
- New layout of SyNCoPy objects on disk: introduction of Spy-containers supporting
multiple datasets/objects within the same folder - First working implementation of
spy.load
andspy.save
- Use dask bags instead of arrays in parallelization engine to permit more
flexible distribution of data across workers - Re-worked
trialdefinition
mechanics: attach the fulltrialdefinition
array
to objects and fetch relevant information on the fly:BaseData._trialdefinition
unifiessampleinfo
,t0
andtrialinfo
and callsdefinetrial
REMOVED
- Removed
dimlabels
property
DEPRECATED
- Retired Dask arrays in
ComputationalRoutine
; use dask bags instead
FIXED
- Flipped sign of offsets in
trialdefinition
to be compatible w/FieldTrip - Enforced PEP8 compliance
- Cleaned up constructor of
BaseData
to prohibit accessing uninitialized attributes
v0.1a0
[v0.1a0] - 2019-07-20
Internal pre-alpha release of SyNCoPy. Prototypes of data format, user-interface
and parallelization framework are in place.
NEW
- Class structure is laid out, meta-functions are present but mostly place-holders
- Support FieldTrip-style calling syntax via
cfg
"structures" (the keys of which are
"unwrapped" by a corresponding decorator) - Preliminary I/O capabilities implemented, objects can be written/read
to/from HDF5 - First prototype of parallelization framework based on Dask
- Custom traceback that is enabeld as soon as SyNCoPy is imported: do not
spill hundreds of lines to STDOUT, instead highlight most probable cause
of error and explain how to get to full traceback if wanted - Basic session management to ensure concurrent SyNCoPy sessions only access
their own data