Closed wbob closed 5 years ago
This is great. I was actually wondering why the repo size had grown so much. I guess I mistakenly committed those a while back and they remained in the git history.
Thanks for the comprehensive issue. I'll run this right now.
Repo size should now be down to a more manageable size. Thanks again!
When cloning the repo on a small bandwith connection I wondered about the relative transfer size vs. the working tree size. When preparing the snap in #22 you must've accidentally added to and then later removed the package blob from the repository. I think there is an argument for a quick clone, even without
--depth 1
. Removing the snap and a python binary reduces the size from 19M to 3M, and there are other artifacts too.I can recommend the oneliner at stackoverflow.com#42544963 to list the biggest blobs and than the java util
bfg
to remove by size or path. It's an easier frontend thangit filter-branch
.Disadvantage of the removal is the rewrite of commit hashes since the introduction of each blob.
bfg
will output a graph for this. People having forked from one of these commits would need to hard reset their forks and rebase. In my opinion this is not an issue with a smaller codebase where most people will prepare a PR from the latest origin/HEAD.Down at 3344K.
optional:
after those deletes:
With all deletes, the repo is down at 540K.