Large files in git about basic-computer-games

[MartinThoma]

The history re-write is done. Please clone from the current repository

[coding-horror]

OK! I cloned the repo from scratch and it looks good to me! Thank you!

[MartinThoma]

If I should do it, I would give @AlaaSarhan time until Monday to finish #404 . If possible, I would like to prevent open PRs.

Should I do it? Does it sound good to wait until #404 is merged / Monday (whatever is earlier)?

[coding-horror]

Sure! Sounds good to me.

[coding-horror]

Yes that sounds good. Let’s re write history and prevent files larger than 1mb.

…

On Fri, Mar 25, 2022 at 3:03 AM Martin Thoma ***@***.***> wrote: I've noticed some astonishingly large files in this repository (in the git history): 81ebba90f411 1,4MiB buildJvm/build_53_King_kotlin/build/install/build_53_King_kotlin/lib/kotlin-stdlib-1.6.0.jar -- added via fe4219d eeb85bc8ea11 4,4MiB 53_King/kotlin/king.jar -- added via fe4219d 7c624e47ff0f 25MiB 89_Tic-Tac-Toe/python/TicTacToe_exe/TicTacToe.exe -- added via 3efe6e3 65c61eb509d6 34MiB 39 Golf/csharp/compiled/linux_x86/golf -- added via 0dbb491 5222483d2f8c 66MiB 39 Golf/csharp/compiled/windows_x86/golf.exe -- added via 0dbb491 They are all already deleted, but still in the git history. This makes this repository way bigger than necessary. I've just seen a warning by Github about which I was confused ... and especially about the executable in 89_Tic-Tac-Toe/python. I initially thought I had done something wrong. Re-write history We could remove those from the git history: https://stackoverflow.com/a/2158271/562769 However, I have to admit that re-writing git history always feels scary to me Prevent it in future I would like to do two things: 1. Add the pre-commit hook check-added-large-files <https://github.com/pre-commit/pre-commit-hooks#check-added-large-files> 2. Check in the CI if a large file was added (the simplest way to do it would be to run pre-commit in CI, but I would like to avoid that ... I need to check how to do this) Does this sound like a good idea to you? What would be a good maximum file size? — Reply to this email directly, view it on GitHub <#681>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALTWVP5KOVCOAXXOYKJZRLVBWFNLANCNFSM5RTWEPNQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[MartinThoma]

Git History Rewrite

I propose to do the following:

1. Clarify if we need the buildJvm directory (discussion) and maybe adjust what gets deleted from the git history
2. Merge all current branches - or ask the people to copy their changes
3. Delete all local branches: git branch -D $(git branch)
4. Create backup: git bundle create backup.bundle --all
5. Rewrite the history (see below - I recommend the automatic way with this list of files to delete)
6. Clean up git repo: git gc --aggressive --prune=now
7. Force-push!!!! Now everybody needs to fork / clone again! ⚠️
8. Ask contributors to (1) delete their fork / repository (2) fork again (3) clone again

Maybe we should also disable force-pushes on main after that in the Github settings?

History Rewrite Option 1: Manual

git filter-branch --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch '39 Golf/csharp/compiled/windows_x86/golf.exe'" \
  --tag-name-filter cat -- --all

git filter-branch -f --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch '39 Golf/csharp/compiled/linux_x86/golf'" \
  --tag-name-filter cat -- --all

git filter-branch -f --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch '89_Tic-Tac-Toe/python/TicTacToe_exe/TicTacToe.exe'" \
  --tag-name-filter cat -- --all

git filter-branch -f --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch '53_King/kotlin/king.jar'" \
  --tag-name-filter cat -- --all

git filter-branch -f --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch 'buildJvm/build_53_King_kotlin/build/install/build_53_King_kotlin/lib/kotlin-stdlib-1.6.0.jar'" \
  --tag-name-filter cat -- --all

git filter-branch -f --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch '08 Batnum/vbnet/.vs/batnum/DesignTimeBuild/.dtbcache.v2'" \
  --tag-name-filter cat -- --all

git filter-branch -f --prune-empty -d /dev/shm/scratch \
  --index-filter "git rm --cached -f --ignore-unmatch '89_Tic-Tac-Toe/python/TicTacToe_exe/assets/tie.png'" \
  --tag-name-filter cat -- --all

History Rewrite Option 2: Automatic

Delete all files from the history that are no longer there. I'm uncertain what happens with files that were moved!

# Install git-filter repo; several options: https://github.com/newren/git-filter-repo/blob/main/INSTALL.md
pip install git-filter-repo

# Run it
git filter-repo --analyze

# Create a list of all files that should get deleted
tail +3 .git/filter-repo/analysis/path-deleted-sizes.txt \
    | tr -s ' ' \
    | cut -d ' ' -f 5- \
    > .git/filter-repo/analysis/path-deleted.txt

# Before you do this, you can check if there are things you want to keep
git filter-repo --invert-paths --paths-from-file .git/filter-repo/analysis/path-deleted.txt --force

rm -rf .git/filter-repo

Alternatively, you could use this path-deleted.txt. I've removed all lines that match

(\.pl|\.py|\.cs|\.java|\.md|\.bas|\.rb|\.gitignore|\.rs|\.js|\.vb|\.sln|\.txt|\.html|\.csproj|\.vbproj|\.kt)$

[MartinThoma]

The effect of this:

$ ls -lh
  74M  backup.bundle
 3,4M  repo-after-history-rewrite-and-clean-manual-deletion.bundle
 3,1M  repo-after-history-rewrite-and-clean-automatic-deletion.bundle

When you now execute this snippet you can see:

2e07bb068f29   41KiB 01_Acey_Ducey/rust/target/debug/incremental/rust-8frg64vi8djd/s-g737sgtzl9-gc3nmb-ydny6jjnqtbz/dep-graph.bin
6557ca9f5546   42KiB 00_Alternate_Languages/88_3-D_Tic-Tac-Toe/csharp/Qubic.cs
595db6616bd6   43KiB 88_3-D_Tic-Tac-Toe/csharp/Qubic.cs
a01944e199e2   43KiB 88_3-D_Tic-Tac-Toe/csharp/Qubic.cs
513651d6a5e4   44KiB 88_3-D_Tic-Tac-Toe/csharp/Qubic.cs
9a4573c62152   48KiB 84_Super_Star_Trek/javascript/superstartrek.mjs
b8ecb9343b12   49KiB 84_Super_Star_Trek/java/SuperStarTrekGame.java
213b2a0b8d38   49KiB 84_Super_Star_Trek/java/SuperStarTrekGame.java
7454180f2ae8   58KiB buildJvm/gradle/wrapper/gradle-wrapper.jar
ea71f1cf7903   58KiB 00_Alternate_Languages/01_Acey_Ducey/elm/package-lock.json
6fe43fa79007  115KiB 75_Roulette/perl/roulette-test.t
d2946a198344  125KiB 00_Alternate_Languages/01_Acey_Ducey/elm/docs/app.js
ccf0500fd14b  196KiB buildJvm/build_53_King_kotlin/build/install/build_53_King_kotlin/lib/kotlin-stdlib-common-1.6.0.jar

In case we don't need the buildJvm folder, we could reduce the allowed file size from 1MB to 200 KB.

[coding-horror]

I think we have definitely decided we don't want any build or IDE specific files in the repo, so proceed accordingly there...