codalab / codabench

Codabench is a flexible, easy-to-use and reproducible benchmarking platform. Check our paper at Patterns Cell Press https://hubs.li/Q01fwRWB0
Apache License 2.0
76 stars 28 forks source link

WIP Downloading submissions v1 #1548

Closed Didayolo closed 4 months ago

Didayolo commented 4 months ago

Original PR

@ mention of reviewers

@Didayolo

Issue resolved

Description

The aim of this PR is to add the feature of downloading all or several submissions at once. This bulk downloading is only accessible for a challenge administrator and is accessible through the submissions manager.

image

image

The user interface view could be improved by placing the drop down menu and apply button on the far right.

This implementation can(/should?) be improved by moving the zip_generator task computation from the django thread to a site-worker by un-commenting # in_memory_zip = stream_batch_download.apply_async((pks,)).get() and commenting in_memory_zip = stream_batch_download(pks) in the file submissions.py line 327.

However, generating zip in a stream by a site-worker is not functional because Celery back end is not configured to return something. This is a part where I would need help if this implementation is mandatory. Also, is there a good way to benchmark the cost of this feature on the Django thread? It should not block the responsiveness of Codabench.

The global implementation follows this path : bulk_download_codabench drawio

A checklist for hand testing

Checklist

Didayolo commented 4 months ago

@nicomy

Tested and it works fine. The only problem is that, when downloading several submissions with the same filenames, they got all zip together and then we can't know which one is which. Maybe we should rename them or add some folder structures.

I merge this PR and keep track of this issue in #1551.