Add command to display the stderr data from submission in terminal

[hkmatsumoto]

The corresponding GCI task

Changes:

• Add a subcommand to show the content of stderr file of the specified submission, invoked by evalai submission SUBMISSION_ID stderr.
• Add a helper function of the subcommand.
• Add tests for the helper function.
• Add an example of the subcommand to documentation.

CC: @vkartik97 utsav @Ram81 @RishabhJain2018

[hkmatsumoto]

Since the content of stderr files are basically error messages, it is hard to distinguish stderr messages (i,e: intended behavior) from cli error messages (i,e: unexpected error) if we print the content directly like the figure below.

To convince users that the output is a stderr message and cli is not crushing, I decided to print an extra message like the figure below.

[hkmatsumoto]

And the content of stderr file is this:

[hkmatsumoto]

Hi @vkartik97, thanks for replying on the task page but I still can't find your review. Could you recomment it please?

[krtkvrm]

Hi,
Can you see this comment: Cloud-CV/EvalAI-ngx#261 (comment)?

[hkmatsumoto]

Yes

[krtkvrm]

The comment is in pending status, so pasting here again:
Why have you made a util: display_submission_stderr, which is only used by stderr, i.e stderr function in submissions.py. I am not able to determine making extracting out the not so large amount statements to a separate util function.
Please let me know if I am missing something.

[hkmatsumoto]

@vkartik97 That is because I tried to make stderr command similar to other functions including submission and result command.
submission calls display_submission_details and result calls display_submission_result.

[hkmatsumoto]

@vkartik97 I personally feel the same thing to you, so I am willing to make another PR that takes in the code of the functions in utils to the command function and wait for opinions from mentors including you.

[krtkvrm]

If you are going ahead for making PR as commented by you, can you only try for stderr(this PR) and result first?

[hkmatsumoto]

Sure but can I ask why?

[krtkvrm]

The refactoring will be less as compared to going for both result and submission. Once we are satisfied with the changes in result, you can go ahead. And pardon me, I didn't state the reason initially.

[hkmatsumoto]

@vkartik97 I got it. Created #253.

[hkmatsumoto]

I fell good about the change made in #253. What do you think @vkartik97?

[hkmatsumoto]

@Ram81 @RishabhJain2018 Could you also take a look?

[hkmatsumoto]

I closed by accident

[hkmatsumoto]

Apart from #253, what do you think of the changes made in this PR?

[krtkvrm]

The PR for #253 seems good, but is of independent of this task. So, approving this, as refactoring can be a new task altogether.