Releases: aazuspan/eerepr
Releases · aazuspan/eerepr
v0.1.2
Changed
eerepr.reset()
now clears the cache immediately instead of waiting foreerepr.initialize()
.
Performance
- Non-deterministic objects are no longer cached, rather than getting a unique attribute to force a cache miss. Negligible reduction in memory, unless you're caching huge shuffled lists.
- The optimization of long truncated lists was modified to check list lengths against a pre-computed max length instead of calculating the minimum string length of each list. Minor speedup when working with a lot of lists.
v0.1.1
Fixed
- Fixed an error #57 generating reprs when the default system encoding is
gbk
Performance
- Avoid stringifying long lists that will definitely be truncated in the repr (~20% speedup when testing with a 25-image Sentinel-2 collection)
Changed
- CSS is loaded from a Python module instead of a static file
v0.1.0
⚠️ Breaking Changes
-
eerepr
no longer initializes on import, and must be manually initialized to work.import eerepr eerepr.initialize()
Added
- Add
on_error
parameter toinitialize
with optionraise
to throw Earth Engine exceptions instead of warning - Add
max_repr_mbs
parameter toinitialize
to allow setting the maximum repr size for safety
Changed
- Pure CSS collapsing (no more JS!)
- Better accessibility - reprs can be navigated by keyboard
- Optimized dict sorting (3-10% faster)
- Improved styling
Fixed
- Replaced deprecated standard lib functions
Removed
- Dropped Python 3.7 support
- Removed automatic
initialize
on import
v0.0.5
v0.0.4
v0.0.3
What's Changed
- Reduce repr storage size by about 30% by @aazuspan in #12
- Available on conda-forge!
Full Changelog: v0.0.2...v0.0.3
v0.0.2
v0.0.1 Initial Release
Update README.md