This is a large, unorganized update driven by specific project needs.
Changes
Removed the dist folder -- no longer supporting easy lambda function setup. This may eventually be done through a Docker file, but in any case we'll make a separate repo for it if there is demand
Features
Improved error message on coord_select node when that node is unable to infer the coordinate step size
Node prober can now handle nodes with multiple values for a single point (i.e. because of an ExpandCoordinates node for statistical operations)
Convolution can now handle requests with dimensions not in the source
Convolution can now handle sources with dimensions not in the kernel_dims
Convolution node can now extend ArrayCoordinates1d to do edges properly -- it does assume a uniform step in the extended direction
Bug Fixes
Added Percentile to list of algorithm classes
There was a tricky floating point error in the signal.py file. When you made an xarray from UniformCoordinates and then re-make the UniformCoordinates from the xarray, the coordinates are mismatched by floating point precision. This causes xarray to no longer match the coordinates -- which causes all kinds of problems.
Convolution now works for ArrayCoordinates1d with only 1 element
This is a large, unorganized update driven by specific project needs.
Changes
dist
folder -- no longer supporting easy lambda function setup. This may eventually be done through a Docker file, but in any case we'll make a separate repo for it if there is demandFeatures
ExpandCoordinates
node for statistical operations)Convolution
can now handle requests with dimensions not in the sourceConvolution
can now handle sources with dimensions not in the kernel_dimsConvolution
node can now extendArrayCoordinates1d
to do edges properly -- it does assume a uniform step in the extended directionBug Fixes
Percentile
to list of algorithm classessignal.py
file. When you made an xarray fromUniformCoordinates
and then re-make theUniformCoordinates
from the xarray, the coordinates are mismatched by floating point precision. This causes xarray to no longer match the coordinates -- which causes all kinds of problems.Convolution
now works forArrayCoordinates1d
with only 1 element