Code repository is almost 100MB big

[kelson42]

Which is a lot, and the content of the history will soon have no value anymore as the dependences will be outdated.

[Jaifroid]

Hmm. Are there tools to reduce size? I guess they mean rewriting history (but if it's not useful history...)?

[Jaifroid]

OK, I believe the reason for this is not the committed code, but the artefacts that are created and archived with each CI run. These artefacts are stored under the respective workflow runs (see example here and look at the bottom of the page under "Artifacts"). These artefacts are kept for 90 days -- see Settings -> Actions -> General. I have been making a lot of runs while creating and debugging the CI. If these artefacts count towards the total reported size, then we should start seeing a size reduction after 90 days. Please note that nightly and release CI runs do not store artefacts in this way. Nightlies are merely uploaded to download.openzim.org, and releases are released in GitHub Releases and on download.openzim.org, but not archived.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be now be reviewed manually. Thank you for your contributions.

[kelson42]

@Jaifroid I just have done it for another repo. Using the given tool is very easy. Let me know if you want me to do it.

[Jaifroid]

@kelson42 I'm very happy for you to have a go. The only BLOBs currently saved to the Repo should be in the Prototype and the Large File test, which we currently still need, though of course there could be more in the (relatively short) history of this Repo from past versions. It's fine to strip out past versions, so long as current files remain. I understand that BFG always preserves the current version of files in each branch, so that shouldn't be an issue (I've used it on Kiwix JS PWA previously, and I'm aware all users need to reset their local copies after the rewrite of history).

I have a plan to move the Prototype to a gh-pages branch which will be built on master and force-pushed each time it is updated, thus ensuring no history is kept on that branch.