nprapps/dailygraphics

Improve graphics archiving process to decrease repo size?

Opened this issue · 1 comments

When archiving by moving directories from the graphics repo into the graphics-archive repo, the filesize of the graphics repo doesn't get trimmed down, since the old visuals are still in the repo's history. For example, you can jump back in time and see all earlier repos using a line like this, which'll display the contents at the end of 2014:

git checkout $(git rev-list -n 1 --before="2014-12-31 23:59" master)

That said, the repo's currently only 50 MB in size as a clone, and GitHub allows repos of up to ~1 GB, so we're on track to break that rule in about a century at this rate, nothing urgent (:

And since the permissions are different between graphics and graphics-archive, (and deleting git history improperly can lead to significant problems,) there may be no reason to change the current workflow. But we could switch to archiving within the graphics repo itself, if filesize had been the main concern of this workflow.

Darn, although now that I look at the uncompressed size of the graphics repo, it's about 450 MB. Not sure which metric GitHub uses to determine if it's "too big to host," but I'd guess it's the less generous version.