Preserving Filenames for Uploads & Results

Working with uploads & results has just gotten easier. Previously, when referencing an uploaded file or result, Nextjournal’s data versioning system would discard the file extension. This was problematic for tools that expect a file to carry a certain extension. This is no longer the case:

artist_data.csv

ls artist_data.csv
 | xargs -n 1 basename

0.9s

Bash

The same is true for results:

echo "my result" > /results/hello.txt

0.2s

Bash

empty

ls hello.txt
 | xargs -n 1 basename

1.4s

Bash

This is all done while preserving the automatic data versioning capabilities that Nextjournal provides.

When working with files that expect a number of files in a certain directory, it's still recommended to create a directory and link the files into it:

mkdir -p /my-data
ln -sf hello.txt
 /my-data/
ln -sf hello.txt
 /my-data/

1.0s

Bash

Then this directory can be read normally:

ls /my-data

0.8s

Bash

Lastly, we've also changed the mount path for the versioned file system from /data to /.nextjournal to avoid collisions when you're working with imported docker images.

We hope these changes remove some of the friction you've previously encountered.