The outline moves, the paths do not

A repository that has promised a journey now spends sixteen days moving the luggage. The subjects get vaguer as the tree gets more organized. Update. test. commit. Clean up. If the history were a diary, these would be the days it forgot to say what happened, and then carefully recorded the furniture.

Scope

This dig examines six commits in bamr87/it-journey from 2020-08-01 through 2020-08-17, in one parent chain after the first dig. It does not include the August 23 fork, pull requests, issues, or any record that a script ran. The evidence index is docs/git-archaeology/001-the-outline-moves/evidence.json in this repository.

Facts

added onedrive and alfred to hb packages modifies init_apps/hb-packages.sh and adds two brew cask install lines, for OneDrive and Alfred. It also adds .DS_Store. The committer is amr abdel-motaleb. git-commit:00dd4b5b4104cfa4f9354e3926672e8e342f41a5

Nine days later, Update deletes zer0_dev.md and init_apps/hb-apps.md, adds init_it.md, and moves several scripts from init_apps/ to init_world/. Git records three of those moves as renames. The new outline still begins # zer0 and still points at ./init_apps/hb-packages.sh and ./init_apps/apm-apps.sh. It also names Level 00 through Level 02, including the heading Level 02 - PIjects. git-commit:ae59de3c5efeaccca16f8d9dc6f06e60a47aa4d8

test adds three files under init_world/hello-gcp/ and modifies the package script, a startup script, and shebang.sh. The subject does not name a test file, and this sample contains no test result. git-commit:a947e6b98f9f0cbb22f18544d1f305b3b78f51f9

commit, four days later, nests scripts into hello-bash, hello-gcp, and hello-mac. Several of those moves are recorded as renames. It also adds iterminate2.sh and vm.yaml, and edits init_it.md. After that edit the outline still contains ./init_apps/hb-packages.sh and ./init_apps/apm-apps.sh, beside a path under init_world/hello-mac/. git-commit:9cc38248d83975bcdd01ab3fa749b07ec66fb52c

Four minutes later, Clean up modifies only init_world/hello-mac/iterminate2.sh. git-commit:0ef2f16839b8ff1421ba3980363da97d9859e09f

Hello win is the first commit in this sample whose author and committer are Bamrman. It adds init_world/hello-gcp/win-sdk.bat, init_world/hello-mac/hello-mac.sh, and init_world/hello-win/hello-win.md. The Windows arrival is a Markdown note plus a batch file. This page does not repeat their install commands. git-commit:7220aefde9b07375a7b64273b4b79f8a2112633b

Inferences

The rename into hello-* folders looks like an attempt to sort the first tools by platform. That reading rests on the paths, not on a stated plan. The outline’s leftover init_apps paths suggest the map was copied forward faster than it was corrected. The objects do not say whether anyone followed those paths and found them missing.

test and commit are subjects with almost no claim. The diffs are doing the talking. That is a habit, not a motive.

Limits

This sample cannot say the packages were installed, the cleanup improved anything, or Windows was set up. It cannot explain the gap from August 1 to August 10. The next commits fork from Clean up’s parent, not from a single line, and they are a different dig. The open question is whether a journey is the outline, or the files the outline can no longer find.

Retrospective

This is hindsight, written now. It is not a motive recovered from the subjects.

I wish the outline had been edited in the same commit that moved the files. Leaving ./init_apps/hb-packages.sh in init_it.md after the rename is a map that sends the next reader to a door the same commit removed. The cheaper habit is: move the file, then fail the commit if the outline still names the old path. git-commit:ae59de3c5efeaccca16f8d9dc6f06e60a47aa4d8

I wish .DS_Store had never entered the tree. A ignore rule on the first day would have saved every later reader from treating desktop debris as project content. git-commit:00dd4b5b4104cfa4f9354e3926672e8e342f41a5

I wish test and commit had been illegal subjects. A subject that names the path and the verb would have saved the time this page spent asking the diff what the author already knew. git-commit:a947e6b98f9f0cbb22f18544d1f305b3b78f51f9