Conversation
* add MNIST and CIFAR * fix CI * fix decoder test * update after ZipArchiveReader changes
* refactor DatasetConfig * add tests
* [PROTOTYPE] add more tests for utils and core API * add pytest-mock as test requirement
* [PROTOTYPE] add more tests for utils and core API * [PROTOTYPE] add caltech256 dataset
* [PROTOTYPE] add support for categories in DatasetInfo * fix setup
| @@ -0,0 +1,5 @@ | |||
| from ._home import * | |||
There was a problem hiding this comment.
I thought asterisk import is prohibited by flake8. Are there such many symbols to be imported here?
There was a problem hiding this comment.
I thought asterisk import is prohibited by
flake8.
Yes, but we blanket ignore F403:
Line 12 in 1140ecf
I'm in the process of fixing that, but I think that is ok in __init__.py files under the condition that the modules define an __all__. This way all the information about what is exposed is contained in the module itself.
Are there such many symbols to be imported here?
Nope, but this way the file stays clean. Following my advice above the namespace also stays clean. The only issue with the style used here is that by looking at the __init__.py we don't know where something is defined.
There was a problem hiding this comment.
After some more thought, I think you are right. We should be explicit there.
|
@mthrok We are going to split this PR up into more reviewable chunks. Do you want me to cc you there? |
sure, thanks. |
No description provided.