which kind of hook do you want to run?
# questionsd
datajoely
02/02/2023, 3:58 PMwhich kind of hook do you want to run?
a
Artur Dobrogowski
02/02/2023, 4:02 PMthe one I've linked - after data catalog creation
Artur Dobrogowski
02/02/2023, 4:03 PMI need to load some items and initialize some singleton class with the data
Artur Dobrogowski
02/02/2023, 4:03 PMand I want it to run only in inference pipeline
d
datajoely
02/02/2023, 4:03 PMI think a
before_pipeline_runa
Artur Dobrogowski
02/02/2023, 4:04 PMyes I'm afraid that before_pipeline_run will be run every time the pipeline is executed
Artur Dobrogowski
02/02/2023, 4:04 PMand I want to package that pipeline as a serving endpoint for mlflow
Artur Dobrogowski
02/02/2023, 4:04 PMand I would like it to run only once for initialization
d
datajoely
02/02/2023, 4:04 PMbut you can check which
--pipelinedatajoely
02/02/2023, 4:04 PMvia
pipeline_namea
Artur Dobrogowski
02/02/2023, 4:05 PMyeah, I'm not familiar enough to get to know where I can get that context in kedro
Artur Dobrogowski
02/02/2023, 4:05 PMso that's why I'm asking where to get that info :)_
Artur Dobrogowski
02/02/2023, 4:06 PMwell I'll do test with before_pipeline as I'm not sure if it will work or not
d
datajoely
02/02/2023, 4:06 PMI’m 99% sure it will work
a
Artur Dobrogowski
02/02/2023, 4:06 PMI mean for it to run only once or many times
d
datajoely
02/02/2023, 4:07 PMso you can find some hacky ways to maintain state like writing a file to disk the 1st time and checking for it’s presence
datajoely
02/02/2023, 4:07 PMa nice way to do this is to create a hook, put a breakpoint in and then play around with the objects live
a
Artur Dobrogowski
02/02/2023, 4:07 PMyeah I often do that by embedding
IPython.embed()2 Views