I’m not really sure thought - I don’t use conda outside of venvs

Hey All! New Kedro user here.. I'm trying to use Kedro test - is there any option to get detailed test summary and assertion errors? I can only see short test summary

What's the naming convention in catalogs when using modular pipelines? In https://github.com/datajoely/modular-spaceflights *

catalog_01_raw.yml

has "companies", "reviews" (no prefixes) *

catalog_02_int.yml

has "data_ingestion.int_typed_companies", "data_ingestion.int_typed_shuttles" (prefix with dot) *

catalog_03_prm.yml

has "prm_shuttle_company_reviews", "prm_spine_table" (prefix with underscore) Or does is it a matter of taste?

kedro-test

Data Catalog

Hi all, I was wondering if anyone knows how I can grab the kedro env flag from within a pipeline.py file, if I run

kedro run --env=test

for instance. I want to pick up the

test

string. Thanks,

So two ways - you can use a

hook

or if you use the

KEDRO_ENV

variable to set it you could just get that within your node

Cool, thanks maybe the environment variable seems like an easier fix!

There is an argument a

before_node_run

hook is the best way to do this, but I think the env variable one is neat

> so how would I get the return of

before_node_run

into a variable. I mostly just use hooks to get the side-effects.

so you could map the

inputs

dict to a new value

but this is more of a philosophical point - we like to think of nodes as pure python functions that have no knowledge of IO operations or execution logic

so if you can do it outside of a node logic it feels more Kedrific

so, the output of

before_node_run

, can be used as an input?

you can mutate the

inputs

provided to the node

not quite sure how I would mutate the inputs from

before_node_run

At this point - maybe we should say the env variable is the simplest solution!

I've made a note to document this as it comes up a lot

one last idea, could you not get the hook to point to a memory_dataset and then read that (in this case the flag for env)

Yeah that's what I was getting to!

nicely put

I ended up doing something slightly amusing. Just combined your ideas used the default project hook to setup the KEDRO_ENV programmatically via python haha.

I like it!

at first I was very confused

maybe we should make that default behaviour 🤔

yea I think that could work quite well.

Hi @User! I was testing this idea but it seams that Hook instances are different when the before_pipeline_run and before_node_run are executed, so the attribute is reset. Is it possible that the Hook class is instantiated twice ? I can see the same attribute in after_catalog_created, before_pipeline_run, but it is None in the berfore_node_run.

hmmm, that's weird. The class is only instantiated once (in

settings.py

) but possibly there's something else going on that's obliterating the attribute. Let me try it out myself and see if it works

Can confirm that it works ok for me. I think you must have some additional logic going on somewhere that is overwriting the attribute somehow?