Simplify downloading of individual collections

[christofs]

Someone interested in just one of the collections still has to download (as ZIP) or clone the entire repository. The repository has more than 500 MB at this point. Is there a way to make it possible for people to download just one given collection?

If we need to split the collections into separate repositories, this would become possible. An advantage would be that each collection would get its own DOI. On the downside, there would be no overarching place anymore, and no joint semantic versioning of releases and DOI would be possible.

[hennyu]

A solution we found for the RIDE repository was to create a separate branch for each issue: https://github.com/i-d-e/ride The data of a branch can be downloaded individually. In the beginning, it was confusing for some because the master branch did not contain anything, but since the issue-branches are linked to from the readme file, the setup is easier to understand. We had the same discussion and the other option would have been to create separate repositories for each issue. I am currently not aware of any other options, but maybe there are?

[ariddell]

You might consider creating and releasing a zip for each directory. On Linux or macOS:

TAG=v.4.0.0
for directory in $(ls -d */ | sed s'#/##'); do
  git archive --format=zip $TAG $directory > $TAG-$directory.zip
done

ergibt

 84M  v.4.0.0-french.zip
 73M  v.4.0.0-italian.zip
113M  v.4.0.0-portuguese.zip
233M  v.4.0.0-spanish.zip

[christofs]

Thanks, Allen, for the idea. But then we have to keep those up-to-date; also, aren't they too big for GitHub (100MB)? So we would have to keep them elsewhere and link to them from Github.

[ariddell]

You would add those zip files to textbox's releases page: https://github.com/cligs/textbox/releases (for the relevant version). The limit is 2G on the files there (source). You would do this every time you tag a new version of textbox.

Don't add them to the repository. You're right, that wouldn't work.

If you made new versions of textbox frequently, and adding the zip files to the releases page became tedious, you could automate the process with travis-ci or some similar service.