-
Notifications
You must be signed in to change notification settings - Fork 134
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Misc fixes/improvements #573
Conversation
So if you used `orient="LR"` it would get lost if you passed in graph_attrs and it didn't have the appropriate value. This fixes that and fixes the unit test and makes sure it's deterministic.
We introspect the dataframe schema a little to capture those aspects. The datatypes are converted to strings to avert any serialization issues.
To have proper project name and version.
Apply Sweep Rules to your PR?
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM -- a few quick fixes
So that people can Try Hamilton :)
Makes it explicit that Hamilton is a first party package.
The problem here was that when we pushed materializers we introduced a regression where somehow input node types got treated differently. The bug here was that "input" nodes are created at add dependency time, thus if two functions request the same input, but provide different types things would correctly error, but in this case in #575 the types were compatible. So the design choice here was to: 1. Check that if one is the subset of the other, then we allow it. 2. We then make the subset type the type of the node. 3. We attach originating functions to ensure we can create a good error message. This has the side-effect of propagating all the way through to Variables and such. 4. The mutating node functions are scoped to only work if the Node is deemed External, that way we don't use that code path inadvertently in the future. 5. We fix an assumption in visualization that assumed inputs didn't have functions. Note: the input/external/user-defined node creation should really be pulled out into a separate step, and not created dynamically in the add_dependency part.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A few nits
So the git commit hook for isort was running with different configuration from the pre-commit CI run. This (a) fixes files to be consistent with (b) the configuration now provided -- e.g. what is first party, and what is local. This should hopefully now prevent any conflicts from reoccurring.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good to me
Fixes/improvees a few things:
Changes
How I tested this
Notes
Checklist