Hi! Before kedro 0.18.0 I was achieving it with the register_config_loader hook because all the params passed through the command line were available in the extra_params. I would also be interested in knowing what is the best practice in the latest version. What I was thinking it may work is providing your custom class for the TemplateConfigLoader intercepting the extra parameters (runtime_params) , updating the global dict and returning the tamplate config loader with the updated global_dict.

Hi, when I try to install kedro 0.17.7, I got an error message indicating that the version does not exist. Is still available via pip? Where do I find all distributions?

It may not work on your version of python. 3.9 support came in 0.18.x

Correct, many thanks @datajoely !

Thanks @PetitLepton @antheas @datajoely @Isaac89 , Extending the TemplateConfigLoader class updating the globals_dict with runtime_params did the trick! 🙏

has anyone used kedro with flask?

can anyone help me pls?

Hey everyone, I'm just wondering what are peoples approaches to using Jupyter Notebooks within Kedro? I believe the general consensus is that Kedro Jupyter Notebooks are still considered to provide use for things like data exploration and prototyping, so the aim is not to for go using them. However, one of the driving forces for me moving to Kedro was the difficulty around version control with Jupyter Notebooks. Is there built in functionality within Kedro for dealing with this? Or are people just excluding notebooks files in .gitignore after conversion of notebooks into pipelines? Interested to get peoples opinions and what is considered to be Kedro best practise regarding this! 🙂

So one of the driving forces for us building kedro originally was to get people out of notebooks when it comes to building robust pipelines

It's a difficult question - personally I commit notebooks in the notebooks directory provided in the template, but they're a communication and prototyping tool, not software in my eyes

I agree, I think they're useful for those reasons, so would like to commit them to the repository! Do you use any tools, like nbconvert or anything to deal with committing notebooks to avoid merge conflicts?

Hi, Question regarding dependencies - would anyone know if fsspec upper boundary is going to be reviewed soon? We are limited to use fsspec >=2022.5.0 so it doesn't play with Kedro 0.18.2. which is limited to <=2022.1.

Can you share why you need a relatively new version of fsspec? Would be great if you can open an GitHub issue about this. We generally don't set upperbound unless there is something breaking. I will have a look today.

There is a chain of dependencies on our side in fact. We need to use package developed by another team that requires s3fs = ">=2022.5.0", which requires mentioned version of fsspec. I read through the history of the fsspec requirement limitation in Kedro repo and seems that there was some breaking changes in the past hence the cautiousness.

Issue opened. Thank you @noklam

Thank you

The python package "nbstripout" works well for me. Install it, add three lines to your .gitattributes file, and it strips all outputs and metadata from a notebook upon "git add *", which makes for a clean git workflow. https://github.com/kynan/nbstripout

hey guys, I was wondering if you guys know any tool that could be used with Kedro that works well obviously, that provides the template for docstrings... I knew some, but they are not up to date anymore...

I'm not sure what you mean - I rely on my IDE configuration to define my doc string templates. Pycharm is best at this, Vs code is ok. For validating format there are also lining tools available

I don't use pycharm, but I mean, there was a tool that was like flake8 that you run it and it does create a docstring template for every function you have in the project...

so that you wouldn't do it manually

I'm not sure if any tools like that. Flake8 will tell you where they're missing but I don't think it will change you code for you. Googling suggests there is a pycharm function to bulk add missing docstrings

nice...I'll take a look at that

thank you

I think in Pycharm if u type """ it should just generate the docstring for you?

On vscode u have autodocstring

There is a setting to do this in bulk apparently https://stackoverflow.com/questions/37549741/is-it-possible-to-bulk-add-docstrings-to-all-the-functions-in-pycharm

Hello guys, I was doing some research about how to run kedro pipelines in containers and I found that the only thing needed besides the dependencies of the project is to have a python virtualenv, am I right or am I missing something?

No that's it! There is also a kedro-docker plugin designed to get you started

Nope, that is one of the great benefits of kerdro being written in python. I started with the kedro docker plugin. That gives you a good starting point for your docker journey, but you might need a few specific things for your environment. Such as system dependencies or database drivers. Once you have on that works for your environment you probably don't need the kedro docker plugin anymore unless you prefer the convenience of kedro cli.