Overview

A feature requested by Paul and Cody, also see here: https://discord.com/channels/862108724948500490/1200119435931942994/1400567949357617302

This allows you to simply specify ucm transcript.in-place -c foo and it will run the transcript on foo in-place.

This uses the default codebase if unspecified, and will create a new codebase, then modify it if passed with -C to a codebase which doesn't already exist.

If the old workaround of transcript.fork -S codebase -C codebase is provided, this will now issue a warning suggesting to use transcript.in-place explicitly instead.

Implementation notes

• Adds a new subcommand
• Adds a new InPlace configuration for the transcript runner
• Refactors the transcript pre-post commands into a bracketed style to ensure resource cleanup
• Detects and warns if using fork with save-to on the same codebase.
• Did a small tweak to normalize output paths, so instead of:

💾  Wrote /Users/cpenner/dev/unison/./gi/basic-transcript.output.md

We now get:

💾  Wrote ./gi/basic-transcript.output.md

If the path is outside of the current dir It will remain an absolute path as you'd expect.

Test coverage

I tested manually:

• Creating a new codebase using transcript.in-place
• Modifying an existing codebase using transcript.in-place
• Modifying my default codebase with transcript.in-place
• Saving a codebase in-place with transcript.fork, and getting a warning. (Also tried with mismatched but canonically equal paths, like relative vs absolute, etc.)
• Tested transcript.fork to ensure it still works
• Tested transcript to ensure it still works