https://dagster.io/ logo
Join Slack
Powered by
Hey there! :wave: This is probably a silly questio...
# ask-community
d
Hey there! 👋 This is probably a silly question but I'm wondering about other people development workflow when working on a Dagster 
@asset
. I realized mine is not very good as I'll work on it mostly on ipython, then copy the code to the 
assets.py
file and run the asset job from Dasgter UI. I'm sure there are better ways (breakpoints, alternatives, ...). Please share!
This came up while working on pulling data from a remote and very slow API. I started to write the asset code and wanted an interactive environment to see the response and parse it to a DataFrame. Ended up copying to a notebook to figure things out iteratively and then back to the 
assets.py
file.
l
First you could start developing in the 
@asset
with the Dagster UI, it's slower but at least you won't have to transition. To skip the UI you could do unit tests and do TDD. Finally, if you really wanted to develop loosely for a long time, you could us https://docs.dagster.io/integrations/dagstermill/using-notebooks-with-dagster
🌈 1
d
Thought on going the Notebook route but was wondering if there are ways to stay in VSCode and use breakpoints or other magic to "pause" the execution and have a live interpreter to debug and edit things
l
Yeah I guess that's not available (I don't use it but was looking into it today for my company)
Seems complicated to me - you'd have to link your juptyer kernel to dagsters job runner.
I think your best bet would be to use the IO_Managers in your notebook to access the desired asset with 
asset_key
d
yep, thats what I'll end up doing
🌈 1
t
If you want to throw a debugger on, this is an example of how you can configure your VS Code to enable that https://github.com/dagster-io/dagster/discussions/10946
d
Ahhh, now we're talking 👌
Will try that tomorrow
3 Views
Open in Slack
PreviousNext
Powered by