packable | Extensive pack and unpack capabilities

This would only work as written if the implementations of Hasher and Packer don't need to know anything about the type they're embedded in (which they won't), in which case they should probably just be plain functions.

It seems like what you really want here is to define methods on State so that the State type implements Hasher and Packer, in which case you should not embed those interfaces. Just define the methods on State correctly.

Interfaces are demonstrated in the Tour of Go.

An infinite loop on git status has been seen before (hbons/SparkleShare issue 1170 as an example)

The next step, with a Git 2.25+, would be to enable trace2, which I present here:

In addition of "How to remove/delete a large file from commit history in Git repository?", the best practice is to use the new tool git filter-repo which replaces BFG and git filter-branch.

It seems there has been previous issues regarding how pandas value_counts() deals with tied values, in an inconsistent way.

As for idxmax() the documentation states clearly:

If multiple values equal the maximum, the first row label with that value is returned.

I am afraid the amount of information you provide is not enough for me to generate a full example with your data but here is an attempt:

EDIT/UPDATE: What actually did the trick for OP was git update-git-for-windows. See comments for details!

Things I'd try in such situations:

• Make sure my current user owns all the current git-internal files, as lacking permissions can interfere with various git operations.
  • Linux: sudo chown -R --changes "$USER" .git
  • Windows: open main folder permissions and let the permissions get rolled down via the Advanced Security Settings dialog (using ☑ Replace all child object permissions with inheritable permissions from this object)
• On Windows: Make sure the path isn't long, as the Windows filesystem NTFS will mess up git if paths get longer than 255 chars/bytes. Keep in mind that the regular git object paths alone (e.g. .git/objects/fe/73f087d8c25fbc1159a50434ef9bf15773e7ac) have a length of 55 characters. Also that special unicode characters can consume more than 1 byte of the file path. Also files in the repo that exceed the maximum path lengths can become problematic for some operations. In worst case just move the repo to e.g. C:\repo\ or something in similar length and try there.
• if even that fails and the repo seems completely broken, I'd try cleaning up the repo by cloning it locally git clone ./repo newrepo and retry the push from within the new repo clone

If you're sure no other Git command is running in that repository at the time you run git gc on that repository, you can add --prune=all. The default is --prune=2.weeks.ago, which gives those other running commands 14 days to complete their work; you can use --prune=1.day.ago to give them less time, for instance.

You can also configure gc.pruneExpire: if not set, it defaults to 2.weeks.ago which is what produces the above default. As j6t notes, the gc.pruneExpire setting variant of this is now rather than all. It's unwise to set now here though: an automatic git gc will use this value and will run in the background, while other Git operations run.

Note that if you have a version of Git >= 2.5 but below 2.15.0, a reduced gc.pruneExpire can break your added work-trees in less time than the default two weeks. The bug is that git gc fails to use the HEAD and index of the added work-tree as a starting point for the reachability traversal of the object DAG. As a result, git gc can remove added blobs that have not yet been committed, and, if you have a worktree with a detached HEAD, even some commits. The best fix for this is to upgrade to 2.15.0 or later, since even the 2-week default isn't necessarily enough.

The test cannot be run due to the folder structure. Your folder structure

Without rewriting history (e.g., rebasing), you cannot remove/alter history. Git by design, uses a chaining of all commits. So, altering an old commit will cause to modify the complete history and all following SHA1s (commit ids). Maybe you still have lots of remote branches on BitBucket (which are not included in your main master branch), Git won't throw away commit if they are referenced from a branch/tag. - For that reason it is important that you force push all branches and (or delete all unured) tags after the rebasing or altering history - otherwise they still point to the old history (which cannot be deleted because some tags/branches points to it).

Big repositories often result from the fact, that the commits contain binaries. if those can be generated using the source, those should not be included in the commits. - If you want to publish them, some platforms provide the option the attach files to tags (oftentimes called releases).

I encountered this exact same error yesterday on macOS when trying to push some binary files to the repo (images, audio). Scripts and other small text files were fine.

After a few hours of hair-pulling, I tried switching the remote URL from HTTPS to SSH and the error is gone.

Before doing that, however, I also tried pushing from a different machine running Windows 10 and there was no error.

How to switch remote URLs from HTTPS to SSH