Diff displayed for 'polish' is too large

[rahimnathwani]

The diff shown below has a change at the start: a two-word phrase replaced with a single word. But because the code is comparing the two drafts as lists of words, each word is now offset by one, making the diff show every single subsequent word as a difference.

[KazKozDev]

Thanks for reporting this! This is a great catch.
You're right - the current word-by-word diff implementation breaks down when the text lengths differ. The offset caused by replacing "sacred covenant" (2 words) with "ritual" (1 word) cascades through the rest of the comparison, making every subsequent word appear as changed even though most of the text is identical.
I'll look into implementing a proper diff algorithm (like Myers' diff or LCS) to handle these cases correctly. The diff should align the unchanged portions properly regardless of length changes.
Appreciate you taking the time to document this with a clear example! 👍