[
https://issues.apache.org/jira/browse/CLIMATE-837?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15401667#comment-15401667
]
Alex Goodman commented on CLIMATE-837:
--------------------------------------
Hi [~huikyole],
Please see my comments on your changes. I think you should make it clear with
what exactly you intended to do with your modifications to my original
DatasetLoader class since it's much different (and heavily stripped down). We
should meet in person ASAP so we are both on the same page with the direction
we intend to go with dataset loading in OCW.
> Make Abstract Base Classes for dataset loaders.
> -----------------------------------------------
>
> Key: CLIMATE-837
> URL: https://issues.apache.org/jira/browse/CLIMATE-837
> Project: Apache Open Climate Workbench
> Issue Type: Improvement
> Components: data sources
> Affects Versions: 1.1.0
> Reporter: Alex Goodman
> Assignee: Huikyo Lee
> Fix For: 1.2.0
>
>
> Now that the {{DatasetLoader}} class has made it possible for users to easily
> handle loading multiple datasets from different sources, [~huikyole] has
> suggested that we add another layer of abstraction one level down.
> Specifically, this pertains to each of the loader functions for each
> different data source (eg, ESGF, OPeNDAP, local, etc). In order to better
> standardize our handling of dataset loaders, we instead propose creating
> abstract base classes for the dataset loaders, as we have done with our
> metrics. For example,
> {code}
> class GenericLoader:
> def load(self, **kwargs):
> # Handle loading
> {code}
> Although the function signatures of each of the source loaders are very
> different, they fundamentally either output a single or list of OCW Dataset
> objects, and may take a single or list of file locations. This would would be
> the basis for defining the abstract base classes.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
