Found in 2.1.28 version

Thanks afs user provided

How can I know the run jupyter file command line in AFS?
Maybe AFS SDK can support it?

Because afs-sdk is part of afs_server's dependency, and afs_server is based on asyncio related modules. And, the most regrettable part is, requests is not support asyncio currently, so, can we switch from requests to aiohttp?

Refactor _create() with *args and **kwargs to make it easier maintenance.

Since the AFS server is already moved to v2 APIs, I think we can deprecate v1 related operations. Add deprecated warnings at this or next version, and really deprecate all v1 operations after the re-implement is done.

Including API are following

• create
• list
• delete

afs2-model is scheduled to on pypi.org 4/29
The reason is WISE-PaaS private pypi server will be shutdown the same day.
Due to many users are develop the old usage on WISE-PaaS, and they will have some checkpoints soon.
The next version 2.1.13 will be just fix the package name, and the usage is the same as before.
I thank @chenjr0719 to give a advanced AFS SDK, and I have already fixed the package name and usage case. It will be appeared in version 2.2. And I will prepared more friendly docs to our users.

As title, the Codecov does not show currently.

Dear Ben:

For pip >= 10, from pip.req import parse_requirements in setup.py will import failed.

Please reference this to make fix this issue.

thanks

How about integrating Travis CI and Codecov to get a better development experience? Here is an example after integrating Travis CI and Codecov.
Once the new PR has been sent, the Travis CI will run unit tests and generate coverage report about this PR. The report will be hosted on Codecov and it also reports the coverage status to the PR. This can help us to check how the PR influence our project. Any idea?

v2.2 has some new documents to user, about RESTful API parameters and stauts code.

• [docs] API return status code
• [docs] model repository API parameters introduction
• [docs] models API parameters introduction

The bottleneck of uploading is about 500-600MB on private cloud.(Maybe less on public cloud)
Currently, the uploading model file is going to api-afs and written to blob.
It may cause timeout when transferring the big file.
I want to add a function to update model to blob directly.

# training... 
# produce model...
...

# afs2-model DO NOT upload 
afs2_model.upload_model(...)

# Model file on blob is models/{instance_id}/{model_respository_id}/{model_id}
# User should input their blob_credential
# Check existed file on the blob and update the big model to the blob directly
# Notify api-afs this operation success/failed
afs2_model.update_model_to_blob(blob_credential, model_file)

There will have v2 AFS API, so SDK have to support v2.
First, we support models.upload_model and the depending method.
I will write the unit test.

The name afs is already registered package name in pypi.
If the project goes on pypi, it will be renamed.

Including API are following

• create
• list
• delete
• tags_value (Not Implement)
• key_list (Not Implement)

@benchuang11046 Notice that we remove all versions of requirements(here). This might cause some issues when using run_jnb because of mismatch version with nbconvert. Maybe we still need to use fixed version of all modules in both jupyter_api_default_requirements.txt and jupyter_app_default_requirements.txt. What's your thought?

How can I Know Models number in AFS?
So, Can you give me the list AFS Models api?

To verify the module work with different versions of python, I want to integrate tox into tests. Also wants to add some lint support in tox.