You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Child of #2799 This is a parent itself and will cover all the changes necessary to popular pages in the "Nodes & Pipelines" section. Won't be addressed until 0.19.0.
The text was updated successfully, but these errors were encountered:
stichbury
changed the title
[Parent] Improve documentation section that covers notes & pipelines
[Parent] Improve documentation section that covers "Nodes & pipelines"
Oct 19, 2023
The imports on top are useful but we should refrain from recommending wildcard * imports.
It describes how to use nodes programmatically, hence how to .run them, how to define their inputs and so on. Most Kedro users will never do this, and only interact with the Framework way of doing things. I think we should consider whether we want to split the page somehow: Framework usage on top, Python API at the bottom. This applies to many other parts of our docs.
I like the use of tiny examples (small functions that only take ‘a’ and ‘b’) to demonstrate the behavior of the nodes. However, the last part on generators completely departs from that and uses a starter, which makes it more complicated but also inconsistent with the rest of the page. I think that example should be remade with tiny functions.
Pipelines:
No imports on top, which would be crucial to know where the pipeline convenience function is imported from.
Speaking of which, some explanation as to what are the differences between the convenience functions (lowercase pipeline) are with respect to the corresponding class initialiser (uppercase Pipeline) would be nice.
Namespaces are mentioned exactly once at the very end of the page, but probably explaining them in more depth would be needed.
My gripes with “The pipeline registry” are mentioned in The pipeline registry is difficult to understand #3233 - this page does some work to explain how to import pipelines, but maybe it should expanded to cover more use cases (even if this technically overlaps with the Pipelines page)
“Run a pipeline” seems to mix content that could make a “Runners” page with several how-to examples, departing from the structure of the previous pages
I don’t understand the “Slice a pipeline” page, it seems to duplicate content from other pages
Child of #2799 This is a parent itself and will cover all the changes necessary to popular pages in the "Nodes & Pipelines" section. Won't be addressed until 0.19.0.
The text was updated successfully, but these errors were encountered: