Few performance updates for Flame, cutting down on time-to run when executing:
re-uses node description now upon repeated re-runs (d5f9f843d628eec0f88d13a1449ee52e4a01e2a4)
drops db only if exists (36e095c8452cdeac1e107560c80adb31abb58377)
most significantly, moves all pipeline imports into __init__.py of entropylab.pipeline instead of entropylab root directory __init__.py, because before running anything under entropylab.flame. ... would unecessarily import all pipeline imports (which are numerous and heavy and were adding significant time to runs) (53e78b9055c6b3d47bda3095e7445a4d68f5349a) -- @qguyk please check this and if happy merge. @galwiner Note that documentation should then be updated as instead of import entropylab now we should use import entropylab.pipeline.
To-do in general: I am wondering if pipeline imports can be better organized so that only necessary modules are imported. Currently import is to wild. :)
Few performance updates for Flame, cutting down on time-to run when executing:
__init__.py
ofentropylab.pipeline
instead ofentropylab
root directory__init__.py
, because before running anything underentropylab.flame. ...
would unecessarily import all pipeline imports (which are numerous and heavy and were adding significant time to runs) (53e78b9055c6b3d47bda3095e7445a4d68f5349a) -- @qguyk please check this and if happy merge. @galwiner Note that documentation should then be updated as instead ofimport entropylab
now we should useimport entropylab.pipeline
.To-do in general: I am wondering if pipeline imports can be better organized so that only necessary modules are imported. Currently import is to wild. :)