As a User, I want to be able to know what a foundry function does without reading docs. I want a clear and intuitive workflow for using datasets
There was some miscommunication in #323 (PR #358 ) about how best to disambiguate load() from load_data(). The solution developed entails privatizing load() by loading upon instantiation of the Foundry class, but this disrupts user workflows for loading multiple datasets, using list() or search(), etc.
The new solution is to break load() out of __init__, but instead disambiguate by renaming and potentially further breaking up the functions. For example, one proposed solution is:
load() --> download_dataset() and view_metadata(), OR view_dataset() where download is an optional boolean parameter
load_data() --> create_dataframe()
Or renaming/refactoring solutions are possible, and require reading through the existing load() and load_data() implementations to decide on next steps
Assumptions:
You are familiar with the contents of load() and load_data() and understand what they currently do
The implementation in #358 is currently in the main branch and needs to change per this ticket
We cannot cut a release unless we either change that or do some git wrangling
You have looked over the Example Workflows defined in our Miro brainstorm
Acceptance Criteria
Given I am a new user of Foundry, I can easily look at the list of functions in my IDE and know exactly what to do to download a dataset, preview the metadata, and use it in a dataframe format
As a User, I can accomplish any of the User Workflows defined without running into friction based on the way downloading, loading, and previewing datasets is defined
### Tasks
- [x ] Review User Workflows (linked above) and discuss user needs and next steps with team
- [ ] Decide on naming solution for `load()` vs `load_data()` and any necessary refactor (see suggested example above)
- [ ] Implement changes
As a User, I want to be able to know what a foundry function does without reading docs. I want a clear and intuitive workflow for using datasets
There was some miscommunication in #323 (PR #358 ) about how best to disambiguate
load()
fromload_data()
. The solution developed entails privatizingload()
by loading upon instantiation of theFoundry
class, but this disrupts user workflows for loading multiple datasets, usinglist()
orsearch()
, etc.The new solution is to break
load()
out of__init__
, but instead disambiguate by renaming and potentially further breaking up the functions. For example, one proposed solution is:load()
-->download_dataset()
andview_metadata()
, ORview_dataset()
wheredownload
is an optional boolean parameterload_data()
-->create_dataframe()
Or renaming/refactoring solutions are possible, and require reading through the existing
load()
andload_data()
implementations to decide on next stepsAssumptions:
load()
andload_data()
and understand what they currently domain
branch and needs to change per this ticketAcceptance Criteria